Society for American Archaeology

Was Agriculture Impossible during the Pleistocene but Mandatory during the Holocene? A Climate Change Hypothesis Author(s): Peter J. Richerson, Robert Boyd, Robert L. Bettinger Source: American Antiquity, Vol. 66, No. 3 (Jul., 2001), pp. 387-411 Published by: Society for American Archaeology Stable URL: http://www.jstor.org/stable/2694241 Accessed: 28/11/2010 19:40

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/action/showPublisher?publisherCode=sam.

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

Society for American Archaeology is collaborating with JSTOR to digitize, preserve and extend access to American Antiquity.

http://www.jstor.org WAS AGRICULTURE IMPOSSIBLE DURING THE PLEISTOCENE BUT MANDATORY DURING THE HOLOCENE? A CLIMATE CHANGE HYPOTHESIS

Peter J. Richerson,Robert Boyd, and RobertL. Bettinger

Several independent trajectories of subsistence intensification, often leading to agriculture, began during the Holocene. No plant-rich intensifications are knownfrom the Pleistocene, even from the late Pleistocene when humanpopulations were oth- erwise quite sophisticated. Recent datafrom ice and ocean-core climate proxies show that last glacial climates were extremely hostile to agriculture-dry, low in atmospheric CO2, and extremnel)'variable on quite short time scales. We hypothesize that agriculturewas impossible under last-glacial conditions. The quite abruptfinal amelioration of the climate wasfollowed imme- diately by the beginnings of plant-intensive resource-use strategies in some areas, although the turn to plants was much later elsewhere. Almost all trajectories of subsistence intensification in the Holocene are progressive, and eventually agriculture became the dominant strategy in all but marginal environments.We hypothesize that, in the Holocene, agriculture was, in the long run, compulsory. We use a mathematical analysis to argue that the rate-limitingprocess for intensification trajectories must generally be the rate of innovation of subsistence technology or subsistence-relatedsocial organization.At the observed rates of innovation, population growth will always be rapid enough to sustain a high level of population pressure. Several processes appear to retard rates of cultural evolution below the maxima we observe in the mostfavorable cases.

Variastrayectorias independenties de la intensificaci6ndel sustento, muchasde las cuales conducierona la agricultura,empezaron duranteel Holoceno. No conocemos ninguna intensificacio que usara muchos recursos vegetales duranteel Pleistoceno, inclu- sive el Pleistoceno iltimo, cuando las poblaciones humanasfueron muy sofisticadas en otros dmbitos.Datos recientes de cilin- dros de hielo sacados de Groenlandia,y de sedimentosocec'nicos, muestran que la l'ltima glaciaci6nfue' extremadamentehostil a la agricultura,ya quefue-seca, baja en C02, y extremadamentevariable en el corto plazo. Proponemosque la agriculturafue imposible en estas condiciones de la l'ltima glaciaci6n. La sabita mejora del clima alfinal de la glaciaci6nfue' seguida inmedi- atamentepor la iniciaci6n de usos mas intensivosde los recursos vegetales en algunos lugares, aunquemucho mas tarde en otras partes. Casi todo las trajectoresde intensificaci6nen el Holoceno eran occurrieronsin retroceso. Finalmente,la agriculturase convirti6en el modoprincipal de sustento en todas partes, exceptopor zonas muyfrias o muy secas. En el Holoceno, hacemos el hip6tesis que la agriculturase vuelve, a final de cuentas, obligatoria. Usamos un andlisis matemdticopara formular el hip6the- sis que los procesos limitantede la taza de intensificaci6ndebe generalmenteser la taza de la inovaci6ntecnol6gica en las estrate- gias de sustento, o la taza de inovaci6n en lasformas de organazaci6nsocial en relaci6n al sustento. Con las tazas de inovaci6n que se observan, el crecimientode la poblaci6n siempre es suficientementerapido como para crear alto nivel de presion pobla- cional. Al parecer, varios processos normalmenteretardan la velocidad de la evoluci6n cultural abajo de las tazas mdximasque observamosen el modelo.

E volutionarythinkers have long been fascinated V. GordonChilde wrote influential books andpapers by the origin of agriculture.Darwin (1874) on the origin of agriculturein the 1920s and 1930s declinedto speculateon agriculturalorigins, (see Flannery1973 andMacNeish 1991:4-19 for the but twentieth-centuryscholars were bolder. The intellectualhistory of the originof agricultureques- Sovietagronomist Nikolai Vavilov, the American geo- tion). These explorationswere necessarilyspecula- grapherCarl 0. Sauer,and the Britisharchaeologist tive andvague, but stimulated interest in thequestion.

Peter J. Richerson * Departmentof EnvironmentalScience and Policy, University of California-Davis, Davis, CA 95616, [email protected] Robert Boyd * Departmentof Anthropology,University of California-LosAngeles, Los Angeles,CA 90095, [email protected] Robert L. Bettinger * Departmentof Anthropology,University of California-Davis,Davis, CA 95616, [email protected] AmericanAntiquity, 66(3), 2001, pp. 387-411 Copyright? 2001 by the Society for AmericanArchaeology

387 388 AMERICAN ANTIQUITY [Vol. 66, No. 3, 2001

Immediatelyafter WorldWar II, the American gathererspersisted to recenttimes in areasseemingly archaeologistRobert Braidwood (Braidwood et al. highly suitablefor agriculture,most notablymuch 1983)pioneered the systematicstudy of agricultural of westernNorth America and Australia.Attempts origins.From the knownantiquity of village sites in to accountfor this rathercomplex pattern are a major the Near East and from the presenceof wild ances- focus of archaeology. tor species of many crops and animaldomesticates in the sameregion, Braidwood inferred that this area Origin of Agriculture as a Natural Experiment was likely a locus of early domestication.He then in Cultural Evolution embarkedon an ambitiousprogram of excavationin The processesinvolved in such a complex phenom- the foothillsof the southernZagros Mountains using enon as the origin of agriculture are many and a multidisciplinaryteam of archaeologists,botanists, denselyentangled. Many authors have given climate zoologists, and earthscientists to extractthe maxi- change a key explanatory role (e.g., Reed mum useful informationfrom the excavations.The 1977:882-883). The coevolutionof humansubsis- availabilityof 14C dating gave his team a powerful tence strategiesand plant and animaldomesticates tool for determiningthe ages of the sites. NearEast- must also play an importantrole (e.g., Blumlerand ern sites older thanabout 15,000 B.P. excavatedby Byrne 1991; Rindos 1984). Hunting-and-gathering Braidwood(Braidwood and Howe 1960) andothers subsistencemay normallybe a superiorstrategy to were occupied by hunter-gathererswho put much incipientagriculture (Cohen and Armelagos 1984; moreemphasis on huntingand unspecializedgath- Harris 1977), and, if so, some local factor may be eringthan on collecting and processingthe seeds of necessaryto provide the initial impetus to heavier especiallyproductive plant resources (Goring-Mor- use of relativelylow-quality, high-processing-effort ris and Belfer-Cohen1998; Henry 1989). Ages are plantresources that eventually result in plantdomes- given here as calendardates before present(B.P.), tication.Population pressure is perhapsthe most pop- where presentis taken to be 1950, estimatedfrom ular candidate(Cohen 1977). Quite plausibly,the 14C dates accordingto Stuiveret al.'s (1998) cali- complex details of local historyentirely determine brationcurves. The Braidwoodteam showed that the evolutionarysequence leading to the origin and about 11,000 years ago, hunter-gathererswere col- spreadof agriculturein everyregion. Indeed, impor- lecting wild seeds, probablythe ancestorsof wheat tantadvances in our understandingof the originsof and barley,and were huntingthe wild ancestorsof agriculturehave resultedfrom pursuitof the histor- domestic goats and sheep. At the 9000 B.P. site of ical detailsof particularcases (Bar-Yosef1998; Flan- Jarmo,the team excavatedan earlyfarming village. nery 1986). Using much the same seed-processingtechnology Nonetheless,we proposethat much about the ori- as theirhunter-gather ancestors 2,000 years before, gin of agriculturecan be understoodin termsof two the Jarmopeople were settledin permanentvillages propositions: cultivatingearly-domesticated varieties of wheatand Agriculture WcasImpossible During The Last barley. Glacial. Duringthe last glacial, climates were vari- Numerous subsequentinvestigations now pro- able andvery dry over largeareas. Atmospheric lev- vide a reasonablydetailed picture of the origins of els of CO2 were low. Probably most important, agriculturein several independentcenters and its last-glacial climates were characterizedby high- subsequentdiffusion to almost all of the earthsuit- amplitudefluctuations on time scales of a decadeor able for cultivation.These investigationshave dis- less to a millennium.Because agriculturalsubsis- covered no region in which agriculturedeveloped tence systems are vulnerableto weatherextremes, earlieror fasterthan in the NearEast, though a North and because the culturalevolution of subsistence Chinesecenter of domesticationof milletmay prove systems making heavy, specialized, use of plant almost as early. Othercenters seem to have devel- resourcesoccurs relatively slowly, agriculture could oped later, or more slowly, or with a different not evolve. sequenceof stages, or all three.The spreadof agri- In The Long Run,Agriculture Is CoinpulsoryIn culturefrom centersof origin to more remoteareas The Holocene. In contrastto the Pleistocene,stable is well documentedfor Europeand NorthAmerica. Holoceneclimates allowed the evolutionof agricul- Ethnographyalso gives us cases wherehunters and turein vast areaswith relativelywarm, wet climates, Richerson et al.] AGRICULTURALORIGINS AND CLIMATE 389 or accessto irrigation.Prehistoric populations tended fall, ice volume, and the like, mostly from cores of to grow rapidlyto the carryingcapacity set by the ocean sediments,lake sediments,and ice caps, pale- environmentand the efficiency of the prevailingsub- oclimatologistshave constructeda stunningpicture sistencesystem. Local communitiesthat discover or of climatedeterioration over the last 14 millionyears acquire more intensive subsistence strategieswill (Bradley1999; Cronin1999; Lamb 1977; Partridge increasein numberand exert competitivepressure et al. 1995). The Earth'smean temperature dropped on smallerpopulations with less intensivestrategies. severaldegrees and the amplitudeof fluctuationsin Thus,in the Holocene,such inter-groupcompetition rainfalland temperatureincreased. For reasonsthat generateda competitiveratchet favoring the origin are as yet ill understood,glaciers wax and wane in and diffusionof agriculture.' concert with changes in ocean circulation,carbon The great variation among local historical dioxide,methane and dust content of theatmosphere, sequencesin the adoptionand diffusionof agricul- and changes in averageprecipitation and the distri- turein theHolocene provides data to testour hypoth- butionof precipitation(Broecker 1995). The result- esis. In the NearEast, agriculture evolved rapidlyin ing patternof fluctuationin climateis very complex. the early Holocene and became a centerfor its dif- As the deteriorationproceeded, different cyclical fusion to the restof westernEurasia. At the opposite patternsof glacial advanceand retreatinvolving all extreme,hunting-and-gathering subsistence systems thesevariables have dominated the pattern. A 21,700- persisted in most of western North America until year cycle dominatedthe early partof the period,a Europeansettlement, despite many ecological sim- 41,000-year cycle between about 3 and 1 million ilaritiesto the NearEast. Thus, each local historical years ago, and a 95,800-year cycle duringthe last sequence is a naturcalexperiment in thefactors that million years (deMenocaland Bloemendal 1995). limitthe rcateof culturalevolution o*f more intensive Milankovich'shypothesis that these variationsare subsistencestrcategies. For our hypothesisto be cor- drivenby changesin the earth'sorbit, and hence the rect, the evolution of subsistencesystems must be solar radiationincome in the differentseasons and rapid compared to the time cognitively modern latitudes, fits the estimated temperaturevariation humanslived under glacial conditions without devel- well, although doubts remain (Cronin 1999: oping agriculture,but slow relative to the climate 185-189). variationthat we proposewas the mainimpediment RcapidClimnate Vcariation in the Late Pleistocene to subsistenceintensification in the late Pleistocene. By culturalevolution we simply mean the change The long-time-scaleclimate change associated with over time in the attitudes,skills, habits,beliefs, and the majorglacial advances and retreats is not directly emotionsthat humans acquire by teachingor imita- relevantto theorigins of agriculturebecause it occurs tion. In our view (Bettinger 1991; Boyd and Rich- so slowly comparedto the rateat which humanpop- erson 1985),culture is best studiedusing Darwinian ulationsadapt by culturalevolution. However, the ice methods.We classify the causes of culturalchange ages also have great variance in climate at much into several"forces." In a very broadsense, we rec- shortertime scales. For the last 400,000 years, very ognize threeclasses of forces: those due to random high-resolutionclimate proxy data are available from effects (the analogs of mutationand drift), natural ice cores taken from the deep ice sheets of Green- selection, and decision-making(invention, individ- landand Antarctica. Resolution of events lastinglit- ual learning,biased imitation, and the like).The deci- tle more thana decade is possible in Greenlandice sion-makingforces will tend to acceleratecultural 80,000 years old, improvingto monthlyresolution evolutionrelative to organicevolution, but by how 3,000 yearsago. Duringthe last glacial, the ice core much is a majorissue in the explanationof agricul- data show that the climate was highly variableon turalorigins. time scales of centuries to millennia (Clark et al. 1999; Dansgaardet al. 1993; Ditlevsenet al. 1996; Was Agriculture Impossible GRIP 1993). Figure 1 shows data from the GRIP in the Pleistocene? Greenlandcore. The 6180 curve is a proxy for tem- The Pleistocenegeological epoch was characterized perature;less negativevalues are warmer.Ca + is a by dramaticglacial advancesand retreats.Using a measureof the amountof dust in the core, which in varietyof proxymeasures of pasttemperature, rain- turn reflects the prevalenceof dust-producingarid 390 AMERICAN ANTIQUITY [Vol. 66, No. 3, 2001

* 900 UnfilteredV18O 34 Ca2+ ~~~~~~~~~~~~~~~~~~~~~~38

600 -42 -46 (ppb) 80 60 40 20

300 150-Year-LowPass Filter -36

-0 -40 -30 -44 18 80 60 40 20

150-Year-HighPass Filter ~~~~~~~~~~~~~~~~~~~~4 (0/00) -11 ~~~~~~~~~~~~~~~~~~~~0 -40 -4 80 60 40 20

| I . I I I I Figure 2. High-resolution analysis of the GRIP ice core 6180 160 120 80 40 0 data by Ditlevsen et al. (1996). The low-pass filtered data 10 Years B.P. shows that the Holocene is much less variable than the Pleistocene on time scales of 150 years and longer. The high- pass filtered data shows that the Pleistocene was also much Figure 1. Profiles of a temperature index, 6180, and an index more variable on time scales less that 150 years. The high- of dust content, Ca2+, from the GRIP Greenland ice core. and low-pass filtering used spectral analytic techniques. 200-year means are plotted. The parts of the GRIP profile These are roughly equivalent to taking a 150-year moving representing the last interglacial may have been affected by average of the data to construct the low-pass filtered series ice flow so their interpretation is uncertain (Johnsen et al. and subtracting the low-pass filtered series from the original 1997). Note the high-amplitude, high-frequency variation in data to obtain the high-pass filtered record. Since layer thin- both the temperature and dust records during the last ning increasingly affects deeper parts of the core by averag- glacial. The Holocene is comparatively much less variable. ing variation on the smallest scales, the high-pass variance is Plotted from original data obtainable at: ftp://ftp. reduced in the older parts of the core. In spite of this effect, ngdc.noaa.gov/paleo/icecore/greenland/summit/ grip/iso- the Pleistocene/Holocene transition is very strongly marked. topes/gripdl8o.txt and ftp:ll ftp.ngdc.noaa.gov/paleo /icecore/greenlandlsummit/grip/chem/ca.txt. andthinning within the ice coreprogressively erases climates. The last glacial period was arid and high-frequencyvariation in the core (visible as the extremelyvariable compared to the Holocene.Sharp narrowingwith increasingage of the 150-yearhigh- millennial-scaleexcursions occur in estimatedtem- pass datain Figure2), the shiftfrom full glacialcon- peratures,atmospheric dust, and greenhousegases. ditions about 18,000 years ago to the Holocene The intensevariability of the last glacialcarries right interglacialis accompaniedby a dramaticreduction down to the limits of the nearly 10-yearresolution in variationon time scales shorterthan 150 years. of the ice core data.The highest resolutionrecords The Holocene (the last relatively warm, ice free in Greenlandice (and lower latituderecords) show 11,600 years) has been a period of very stable cli- that millennial-scalewarmings and coolings often mate, at least by the standardsof the last glacial.2 beganand ended very abruptlyand were oftenpunc- The climate fluctuationsrecorded in high-lati- tuatedby quite large spikes of relativewarmth and tude ice cores are also recordedat latitudeswhere cold with durationsof a decadeor two (e.g., Grafen- agriculture occurs today. Sediments overlain by stein et al. 1999). Figure 2 shows Ditlevsen et al.'s anoxic waterthat inhibits sediment mixing by bur- (1996) analysis of a Greenlandice core. Not only rowing organismsare a sourceof low- and mid-lat- was the last glacial much more variable on time itudedata with a resolutionrivaling ice cores.Events scales of a century-and-a-halfor more (150-year recordedin NorthAtlanticsediment cores are closely low-passfilter) but also on muchshorter time scales coupledto those recordedin Greenlandice (Bond et (150-year high-pass filter). Even though diffusion al. 1993), but so arerecords distant from Greenland. Richerson et al.] AGRICULTURALORIGINS AND CLIMATE 391

Hendy and Kennett(2000) reporton watertemper- where proxy data often do not show a cold period atureproxies from sediment cores from the often- coincidingwith the YoungerDryas, althoughsome anoxic SantaBarbara Basin just offshoreof central recordsshow a similarAntarctic Cold Reversaljust California.This datashows millennial-and sub-mil- antedatingthe NorthernHemisphere Younger Dryas lennial-scaletemperature fluctuations from 60-18 (Bennettet al. 2000). thousandyears ago with an amplitudeof about8?C, Otherrecords provide support for millennial-scale compared to fluctuations of about 2?C in the climatefluctuations during the last glacial thatcan- Holocene.As in the Greenlandcores, the millennial- not be convincinglycorrelated with the Greenland scale events often show very abruptonsets and ter- ice record.Cronin (1999:221-236) reviews records minationsand are often punctuatedby brief spikes fromthe deep tropical Atlantic, Western North Amer- of warmthand cold. Schulz et al. (1998) analyzed ica, Florida,China, and New Zealand.Recent notable organicmatter concentrations in sedimentcores at additionsto his cataloginclude southern Africa (Shi oxygen minimum depths from the Arabian Sea et al. 2000), the AmericanMidwest (Dorale et al. depositedover the past 110 thousandyears. The vari- 1998), the Himalayas (Richardset al. 2000), and ationin organicmatter deposited is thoughtto reflect northeasternBrazil (Behling et al. 2000). Clapper- the strengthof upwelling,driven by changes in the ton (2000) gives evidencefor millennial-scale glacial strengthof the ArabianSea monsoon.AMS '4C dat- advances and retreatsfrom most of the American ing of boththe Arabian Sea andSanta Barbara cores cordillera-Alaska and western North America gives good time control in the upper part of the throughtropical America to the southernAndes. record,and the climate proxy variationis easily fit While the complexfeedback processes operating to Greenlandice millennial-scaleinterstadial-stadial in the atmosphere-biosphere-oceansystem are not oscillations.Allen et al. (2000) examine the pollen completely understood(Broecker 1995:241-270), profiles from the laminated sediments of Lago plausible physical mechanismscould have linked Grandedi Monticcio in southernItaly. Changes in temperaturefluctuation in theboth hemispheres. For the proportionof woody taxa in the core were dom- example,Broecker and Denton (1989) proposedan inatedby large-amplitudechanges near the limits of explanationbased upon the effects of glacial melt- resolutionof the data, abouta century.The millen- wateron the deep circulationof the NorthAtlantic. nial-scalevariations in this core also correlatewith Today,cold, saltywater from the surfaceof the North the Greenlandrecord. Peterson et al. (2000) show Atlantic is the source of about half of the global thatproxies for the tropical Atlantic hydrologic cycle ocean'sdeep water.This largeoutflow of deep water have a strong millennial-scalesignal that likewise currentlymust be balancedby an equallyenormous closely matchesthe Greenlandpattern. inflow of warmsurface and intermediatewater into Reportsof proxyrecords apparently showing the the high NorthAtlantic. If glacial melt water low- ultimate Younger Dryas millennial-scale cold eredthe salinityof the NorthAtlantic and interrupted episode, strongly expressed in the North Atlantic the flow of deep water,the whole coupled atmos- records12,600-11,600 B.P., have been reported from phere-oceancirculation system of the world would all over the world,including southern German oxy- be perturbed.Broecker and Denton's hypothesis gen isotope variations (Grafenstein et al. 1999), explainshow the northernand southern Hemisphere organic geochemistry of the Cariaco Basin, temperatureand ice fluctuationscould have been in Venezuela(Werne et al. 2000), New Zealandpollen phaseeven thoughthe directeffects of orbital-scale (Newnhamand Lowe 2000), and Californiapoller variationon the two hemispheresare out of phase. (West 2000). The Younger Dryas episode has Impactsof Millennial-and Sub-Millennial-Scale receiveddisproportionate attention because the timc Variationon Agriculture periodis easily datedby 14C andis sampledby many lake and mountainglacier cores too short to react We believe thathigh-frequency climate and weather older millennial-scale events. As Cronin (1999 variationwould have made the evolution of meth- 202-221) notes, the YoungerDryas is frequently ods for intensive exploitation of plant foods detectedin a diversearray of NorthernHemisphere extremelydifficult. Holocene weather extremes sig- climateproxies from all latitudes.The maincontro. nificantly affect agricultural production (Lamb versy involvesdata from the SouthernHemisphere 1977). Forexample, the impactof the LittleIce Age 392 AMERICAN ANTIQUITY [Vol. 66, No. 3, 2001

(400-150 B.P.) on Europeanagriculture was quite ther field division would have been successful at significant(Grove 1988). The Little Ice Age is rep- coping with muchlarger amplitude fluctuations that resentativeof the Holocene millennial-scalevaria- occurred during the last glacial. We expect that tion thatis very much more mutedthan last-glacial opportunismwas the most importantstrategy for eventsof similarduration. Extreme years during the managingthe risks associatedwith plantfoods dur- Little Ice Age caused notable famines and such ing the last glacial.Annual plants have dormant seed extremes would have been more exaggeratedand thatspreads their risk of failureover many years, and morefrequent during last glacial times. The United perennialsvary seed output or storage organ size NationsFood andAgriculture Organization's (2000) substantiallybetween years as weatherdictates. In a Global Informationand EarlyWarning System on highly variable climate, the specialization of Food andAgriculture gives a usefulqualitative sense exploitationon one or a few especially promising for the currentimpacts of inter-annualweather vari- specieswould be highlyunlikely, because "promise" ation on food production.Quantitative estimates of in one year or even for a decade or two would turn currentcrop losses due to weathervariation are dif- to runsof years with little or no success. However, ficult to make,but reasonableestimates run 10 per- most years would likely be favorable for some cent on a country-widebasis (Gommes 1999) and species or another,so generalizedplant-exploitation perhaps10-40 percenton a state basis in Mexico, systemsare compatible with highly variable climates. depending upon mean rainfall (Eakin 2000). The acorn-relianthunter-gatherers of California,for Gommesbelieves that weather problems account for example, used several kinds of oak, gatheringless half of all crop losses. favored species when more favored ones failed If losses in the Holoceneare this high andif high- (Baumhoff 1963:Table 2). Reliance on acorns frequency climate variation in the last glacial demandedthis generalizedpattern of species diver- increasedat lower latitudesroughly as much as at sificationbecause the annualproduction of individ- Greenland,a hypotheticallast-glacial farming sys- ual trees is highly variablefrom year to year,being tem would face cripplinglosses in more years than correlatedwithin species but independentbetween not. Devastatingfloods, droughts,windstorms, and species(Koenig et al. 1994).Pleistocene hunter-gath- otherclimate extremes,which we experienceonce erersystems must have been even morediversified, a century,might have occurred once a decade.In the lackingthe kindof commitmentto a single resource tropics,rainfall was highlyvariable (Broecker 1996). category(acorns) observed in California. Few yearswould be suitablefor good growthof any The evolutionof intensiveresource-use systems given plantpopulation. Even under relatively benign like agricultureis a relativelyslow process, as we Holocene conditions agriculturalistsand intensive documentbelow. If ecologicaltime-scale risks could plant collectors have to make use of risk-manage- be managedsome way,or if some regionslacked the mentstrategies to cope with yield variation.Winter- high-frequencyvariation detected by the as yet few halder and Goland (1997) use optimal foraging high-resolutionclimate proxy records, the evolution analysisto arguethat the shiftfrom foraging to agri- of sophisticatedintensive strategieswould still be culturewould have requireda substantialshift from handicappedby millennial-scalevariation. Plant and minimizingrisk by sharingto minimizing risk by animalpopulations responded to climaticchange by field dispersal.Some ethnographicallyknown East- dramaticallyshifting their ranges, but climate change ernWoodland societies that mixed farming and hunt- was significanton the time scales shorterthan those ing, forexample the Huron,seemed not to havemade necessaryfor rangeshifts to occur.As a result,last- this transitionand to have suffered frequentcata- glacial naturalcommunities must have always been strophicfood shortages.Storage by intensiveplani in the process of chaotic reorganizationas the cli- collectorsand farmers is anexcellent means of meet- matevaried more rapidly than they could reach equi- ing seasonal shortfalls,but is a marginalmeans o1 librium.The pollen recordfrom the Mediterranean coping with inter-annualrisk, much less multi-yeaw and Californiaillustrates how much more dynamic shortfalls(Belovsky 1987:60).3 plantcommunities were during the lastglacial (Allen If Winterhalderand Golandare correctthat con. et al. 1999; Heusser1995). Pleistocenefossil beetle siderable field dispersal is required to manage faunaschange even morerapidly than plants because Holocene yield risks, it is hardto imaginethat fur, manyspecies, especially generalist predators, change Richerson et al.] AGRICULTURALORIGINS AND CLIMATE 393 theirranges more rapidly than plants. Hence they are 300 - better indicatorsof the ecological impacts of the A abrupt,large-amplitude climate changes recorded by the physical climate proxies from the last glacial (Coope 1987). 250 - Could the evolution of intensiveplant-exploita- tion systems have trackedintense millennial- and submillennial-scalevariation? Plant food-rich diets take considerabletime to develop. Plant foods are 200 / generally low in proteinand often high in toxins. Some time is requiredto work out a balanceddiet AtmosDheric rich in plant foods, for example, by incorporating CO2 pmol/mol legumesto replacepart of the meatin diets.Whether 150- l l intensificationand agriculturealways lead to health 60 30 0 declines due to nutritionalinadequacy is debatable, butthe potentialfor them to do so absentsometimes- 10 Years B.P. subtle adaptationsis clear (Cohen and Armelagos 1984;Katz et al. 1974).The seasonalround of activ- ities has to be much modified, and women's cus- 100% B tomaryactivities have to be given moreprominence . 0 relativeto men's hunting.Changes in social organi- 0. ** zationeither by evolutionin situor by borrowingtend 0 to be slow (Bettingerand Baumhoff 1982; North and *0. Thomas 1973). We doubt that even sophisticated 0e ** last-glacialhunter-gatherers would have been able to solve thecomplex nutritional and scheduling prob- Biomass or Seed lems associatedwith a plant-richdiet while coping Yield Relative to . Current Level with unpredictablehigh-amplitude change on time 0%- I scales shorterthan the equilibrationtime of plant 100 200 300 400 migrationsand shorterthan actual Holocene trajec- Ambient C02 pmollmol tories of intensification.In keeping with our argu- ment, the direct archaeologicalevidence suggests Figure 3. Panel A shows the curve of atmospheric CO2 as thatpeople began to use intensivelythe technologies estimated from gas bubbles trapped in Antarctic glacial ice. Data from Barnola et al. (1987). Panel B summarizes thatunderpinned agriculture only afterabout 15,000 responses of several plant species to experimental atmos- B.P. (Bettinger2000). pheres containing various levels of CO2* Based on data sum- marized by Sage (1995). Carbon Dioxide Limitation of Photosynthesis Plantproductivity was also limitedby lower atmos- per unit CO2fixed, exacerbatingthe ariditycharac- phericCO2 during the last glacial.The CO2content teristic of glacial times. Beerling (1999) estimates of the atmospherewas about190 ppmduring the last the totalorganic carbon stored on land as a resultof glacial,compared to about250 ppmat the beginning photosynthesisduring the Last Glacial Maximum of the Holocene(Figure 3). Photosynthesison earth using a spatiallydisaggregated terrestrial plant pro- is CO2-limited over this rangeof variation(Cowling duction model coupled to two differentglobal cli- and Sykes 1999; Sage 1995). Beerling and Wood- mate models to providethe environmentalforcing ward(1 993; see also Beerlinget al. 1993)have shown for plantgrowth. The model resultsdiffer substan- that fossil leaves from the last glacial have higher tially,one indicatinga 33 percentlower, and the other stomataldensity, a featurethat allows higher rates of a 60 percentlower, terrestrial carbon store at the Last gas exchange needed to acquireCO2 under more GlacialMaximum compared to the Holocene.Mass- limiting conditions. This higher stomatalconduc- balance calculationsbased on stable isotope geo- tance also causes higher transpirationwater losses chemistryalso indicate a qualitativelylarge drop, 394 AMERICAN ANTIQUITY [Vol. 66, No. 3, 2001 but uncertaintiesregarding terrestrial 613C lead to a dantplant resources more efficiently,but only, cur- similarlylarge range of estimates. Low mean pro- rent evidence suggests, duringthe B0lling-Aller0d ductivity,along with greatervariance in productiv- periodof near-interglacialwarmth and stability. The ity, would have greatlydecreased the attractiveness Natufiansequence in the Levantis the best-studied of plantresources during the last glacial. andso farearliest example (e.g., Bar-Yosefand Valla Lower averagerainfall and carbondioxide dur- 1991). One last siege of glacialclimate, the Younger ing the lastglacial reduced the areaof theearth's sur- Dryasfrom 12,900 B.P.until 1 1,600 B.P.,reversed face suitable for agriculture (Beerling 1999). these trendsduring the Late Natufian(e.g., Goring- Diamond(1997) arguesthat the rateof culturalevo- Morrisand Belfer-Cohen 1998). The Younger Dryas lutionis morerapid when innovationsin local areas climatewas appreciablymore variable than the pre- can be sharedby diffusion.Thus, a reductionin the cedingAller0d-B0lling and the succeedingHolocene areasuitable for agricultureand the isolationof suit- (Grafensteinet al. 1999;Mayewski et al. 1993).The able areasfrom one anotherwill have a tendencyto ten abrupt,short, warm-coldcycles that punctuate reducethe rateof intensificationand make the evo- the YoungerDryas ice recordwere perhapsfelt as lution of agricultureless likely in any given unit of dramaticclimate shifts all aroundthe world. After time. Since the slowest observedrates of intensifi- 11,600 B.P.,the Holoceneperiod of relativelywarm, cation in the Holocenefailed to resultin agriculture wet, stable, CO2-richenvironments began. Subsis- untilthe Europeaninvasions of the last few hundred tence intensificationand eventuallyagriculture fol- years,a sufficientslowing of the rateof evolutionof lowed. Thus, while not perfectlyinstantaneous, the subsistencecould conceivablyin itself explain the shift from glacial to Holocene climates was a very failureof agricultureto emergebefore the Holocene. largechange, and took placemuch more rapidly than A slower rateof culturalevolution would also tend culturalevolution could track. to preventthe rapid adaptation of intensivestrategies Mightwe not expect agricultureto haveemerged duringany favorablelocales or periods that might in the last interglacial130,000 years ago or even dur- have existed duringthe last glacial. ing one of the even older interglacials?No archaeo- On present evidence we cannot determine logical evidence has come to light suggesting the whetheraridity, low CO2levels, millennial-scalecli- presenceof technologiesthat might be expectedto mate variability,or sub-millennial-scale weather accompanyforays into intensiveplant collecting or variationwas the mainculprit in preventingthe evo- agricultureat this time. Anatomically modern humans lutionof agriculture.Low CO2and climate variation mayhave appeared in Africa as earlyas 130,000years wouldhandicap the evolution of dependenceon plant ago (Klein 1999: Ch.7), but they were not behav- foods everywhereand were surely more significant iorallymodern. Humans of the last interglacialwere thanbehavioral or technological obstacles. Hominids uniformlyarchaic in behavior.Very likely, then, the evolvedas plant-usingomnivores (Milton 2000), and humansof the last interglacialwere neithercogni- the basic technologyfor plantexploitation existed at tively norculturally capable of evolving agricultural least ten thousandyears before the Holocene (Bar- subsistence.However, climate might also explainthe Yosef 1998). At least in favorablelocalities, appre- lackof markedsubsistence intensification during pre- ciable use seems to have been made of plantfoods, vious interglacials.Ice cores from the thickAntarc- includinglarge-seeded grasses, well back into the tic ice cap at Vostokshow thateach of the last four Pleistocene (Kislev et al. 1992). Significantly,we interglacialsover the last 420,000 yearswas charac- believe,the use of suchtechnology over spans of last- terizedby a short,sharp peak of warmth,rather than glacialtime that were sufficient for successivewaves the 11,600-year-long stable plateau of the Holocene of intensificationof subsistencein the Holocene led (Petit et al. 1999). Further,the GRIP ice core sug- to only minorsubsistence intensification, compared gests the last interglacial(1 30,000-80,000 B.P.)was to the Mesolithic, Neolithic, and their ever-more- morevariable than the Holocene,although its lackof intensivesuccessors. agreementwith a nearbyreplicate core for this time periodmakes this interpretationtenuous (Johnsen et Subsistence to Amelioration Responses al. 1997). On the other hand,the atmosphericcon- As the climateameliorated, hunter-gatherers in sev- centrationof CO2was higher in the three previous eralparts of the worldbegan to exploitlocally abun- interglacialsthan during the Holocene,and was sta- Richerson et al.] AGRICULTURALORIGINS AND CLIMATE 395 ble at high levels for about 20,000 years following moreland-efficient subsistence systems lead to pop- the warmpeak during the lastinterglacial. The highly ulation growth and labor intensification.Locally, continentalVostok site unfortunatelydoes not record hunter-gatherersmay win some battles(e.g., in the the same high-frequencyvariation in the climateas GreatBasin, Madsen 1994), but in the long run the most otherproxy climaterecords, even those in the more intensive strategieswill win whereverenvi- southernhemisphere (Steig et al. 1998).Some north- ronmentsare suitablefor theirdeployment. ernhemisphere marine and terrestrial records suggest Thearchaeology supports this argument (Bettinger that the last interglacialwas highly variablewhile 2000). Societies in all regionsof the world undergo otherdata suggest a Holocene-lengthperiod of sta- a very similar pattern of subsistence efficiency ble climatesca. 127,000-117,000 B.P.(Frogley et al. increase and populationincrease in the Holocene, 1999). Betterdata on the high-frequencypart of the albeit at very differentrates. Holocene hunter-gath- Pleistocenebeyond the reach of the Greenlandice erers developed local equilibriathat, while some- cores is neededto test hypothesesabout events ante- times lasting for thousandsof years, were almost datingthe latest Pleistocene. Long marine cores from alwaysreplaced by more intensiveequilibria. areasof rapidsediment accumulation are beginning to revealthe millennial-scalerecord from previous Alternative Hypotheses Are Weak glacial-interglacialcycles (McManuset al. 1999).At Aside from other forms of the climate-change least the last five glacialshave millennial-scalevari- hypotheses described above, archaeologists have ationsmuch like the last glacial.The degreeof fluc- proposed three prominent hypotheses-climate tuationsduring previous interglacials is still notclear, stress, populationgrowth, and culturalevolution- butat least some proxy data suggest that the Holocene to explain the timing of agriculturalorigins. They has been less variablethan earlier interglacials (Poli were formulated before the nature of the Pleis- et al. 2000). tocene-Holocenetransition was understood,but are still the hypotheses most widely entertained by During The Holocene, Was Agriculture archaeologists(MacNeish 1991). None of the three Compulsory In The Long Run? provides a close fit with the empiricalevidence or Once a more productivesubsistence system is pos- to theory. sible, it will, over the long run,replace the less-pro- ClimateStress WasFirst Too ductive subsistence system that preceded it. The Common, ThenToo reasonis simple:all else being equal,any groupthat Rare can use a tractof land more efficientlywill be able Childe (1951) proposed that terminalPleistocene to evict residentsthat use it less efficiently(Boserup desiccationstressed forager populations and led to 1981; Sahlinsand Service 1960:75-87). More pro- agriculture.Wright (1977) arguedthat Holocene cli- ductiveuses supporthigher population densities, or mate ameliorationbrought pre-adapted plants into morewealth per capita, or both. An agriculturalfron- the Fertile Crescent areas where agriculturefirst tierwill tendto expandat the expense of hunter-gath- evolved. Bar-Yosef(1998) and Moore and Hillman erers as rising populationdensities on the farming (1992) argue that Late Natufiansedentary hunter- side of the frontiermotivate pioneers to invest in gatherersprobably undertook the first experiments acquiringland from less-efficient users. Farmers may in cultivationunder the pressure of theYoungerDryas offerhunter-gatherers an attractivepurchase price, a climatedeterioration. Natufian peoples lived in set- compellingidea about how to becomericher through tled villages and exploited the wild ancestors of farming,or a dismalchoice of flight, submission,or wheat and barleybeginning in the Aller0d-B0lling militarydefense at long odds againsta morenumer- warm period (14,500-12,900 B.P.) (Henry 1989), ous foe. Early farmers(and other intensifiersmore and then revertedto mobile hunting-and-gathering generally)are also liable to targetopportunistically duringthe sharp,short Younger Dryas climate dete- high-rankedgame and plant resourcesessential to rioration(12,600-11,600 B.P.), the last of the high- their less-intensive neighbors, exerting scramble amplitudefluctuations that were characteristic of the competitivepressure on themeven in the absenceof last glacial (Bar-Yosefand Meadow 1995; Goring- aggressive measures. Thus, subsistence improve- Morrisand Belfer-Cohen1997). Post-Natufiancul- mentgenerates a competitiveratchet as successively turesbegan to domesticatethe same species as warm 396 AMERICAN ANTIQUITY [Vol. 66, No. 3, 2001 and stable conditions returnedafter the Younger populationgradually drove people to intensifysub- Dryas,around 1 1,600 B.P. Unfortunately,a flat spot sistencesystems to relieveshortages caused by pop- in the 14C/calendar-yearcalibration curve makes pre- ulation growth, eventually triggering a move to cise datingdifficult for the most criticalseveral hun- domesticates.Looked at one way, populationpres- dredyears centered on 11,600 B.P.(Fiedel 1999).As sure is just the populationgrowth part of the com- a componentof an explanationof a local sequence petitive ratchet. However, this argument fails to of change,such hypotheses may well be correct.Yet, explain why pre-agriculturalhunter-gatherer inten- theybeg thequestion of why the 15 or so similardete- sificationand the transitionto agriculturebegan in riorationsand ameliorationsof the last glacial did numerouslocations after 11,600 yearsago (Hayden not anywherelead to agricultureor why most of the 1995).Assuming that humans were essentially mod- later origins of agricultureoccurred in the absence em by the UpperPaleolithic, they would have had of YoungerDryas-scale deteriorations. Note alsothat, 30,000 years to build up a populationnecessary to in principle, populationscan adjust downwardto generatepressures for intensification. Given any rea- lower carryingcapacities through famine mortality sonableestimate of the humanintrinsic rate of nat- even more quickly thanthey can grow up to higher uralincrease under hunting-and-gathering conditions ones. Such hypothesescannot, we believe, explain (somewhatless than 1%yr-1 to 3%yr-1), populations the longertime- and largerspatial-scale problem of substantiallybelow carTyingcapacity will doublein the absenceof agriculturein the Pleistoceneand its a centuryor less, as we will see in the models that multipleorigins and rapidspreads in the Holocene. follow. The details of subsistence responses to the A Basic Modelof PopulationPressure. Since the YoungerDryas in the areasof early origins of agri- population explanation for agricultureand other culturewill eventuallyproduce a sharptest of the adaptivechanges4 connected with increasedsubsis- variabilityhypothesis. We suggestthat the lateNatu- tenceefficiency remains very popular among archae- fian de-intensificationin response to the Younger ologists, we take the time here to examine its Dryas was a retreatfrom the trendleading to agri- weaknessformally. The logisticequation is one sim- cultureand was unlikelyto have producedthe first ple, widelyused model of thepopulation growth. The steps toward domestication.More likely, the late rateof changeof populationdensity, N, is given by: Natufianpreserved remnants of earlier,more inten- sive Natufiantechnology and social organization that dN ( N (1) the -=rN(1--K) N servedto start Levantinetransition to agriculture dt KK at an unusually advancedstage after the Younger Dryas ended. Events in the Younger Dryas time where r is the "intrinsicrate of naturalincrease"- periodalso providean opportunityto investigatethe the rateof growthof populationdensity when there effectsof CO2concentration partly independently of is no scarcity-and K is the "carryingcapacity," the climatevariability. The rise in CO2concentration in equilibrium population density when population the atmospherebegan two to threemillennia before growthis haltedby density-dependentchecks. In the temperaturesbegan to rise andcontinued to increase logistic equation,the level of populationpressure is steadily throughthe YoungerDryas (Sowers and given by the ratio N/K. When this ratio is equal to Bender 1995). The YoungerDryas periodde-inten- zerothe population grows at its maximumrate; there sificationof the Natufiansuggests an independent is no populationpressure. When the ratiois one, den- effect of millennialand/or sub-millennial variability. sity dependenceprevents any populationgrowth at all. It is easy to solve this equationand calculate the PopulationGrowth Has the WrongTime Scale lengthof time necessaryto achieveany level of pop- Cohen's (1977) influentialbook arguedthat slowly ulationpressure, t = N/K. accumulatingglobal-scale population pressure was responsiblefor the eventual origins of agriculture i beginningat the 11,600 B.P. time horizon.He imag- T(7z) rlni( IIo (2) ines, quite plausibly,that subsistenceinnovation is driven by increases in population density, but, where it0 is the initial level of populationpressure. implausiblywe believe, thata long, slow buildupof Let us very conservativelyassume that the initial Richerson et al.] AGRICULTURALORIGINS AND CLIMATE 397 populationdensity is only 1 percentof what could be sustainedwith the use of simple agriculture,and dN(x) N(xjIl N(x) + d2 d2N(x) 3 thatthe maximumrate of increaseof humanpopu- at K J dx2 lationsunconstrained by resourcelimitation is 1 per- popuilationi growth dispersal cent per year. Under these assumptions, the Here N(x) is the populationdensity at a pointx in a populationwill reach 99 percentof the maximum one-dimensionalenvironment. Equation (3) says that populationpressure (i.e., 7t = .99) in only about920 the rateof change of populationdensity in a partic- years. Serendipitousinventions (e.g., the bow and ularplace is equalto the populationgrowth there plus arrow)that increase carrying capacity do not funda- the net effect of random,density-independent dis- mentallyalter this result.For example, only the rare persal into and out of the region. The parameterd single inventionis likely to so much as double car- measuresthe rate of dispersaland is equalto the stan- ryingcapacity. If such an inventionspreads within a darddeviation of the distributionof individualdis- populationthat is near its previouscarrying capac- persal distances. In an environmentthat is large ity, it will still face half the maximumpopulation comparedto d, a small populationrapidly grows to pressureand thus significant incentive for further nearcarrying capacity at its initiallocation, and then, innovation.At an r of 1 percentsuch an innovating as shownin Figure4 (redrawnfrom Ammerman and populationwill again reach99 percentof the maxi- Cavalli-Sforza1984), begins to spreadin a wave-like mum populationpressure in 459 years. fashion across the environmentat a constantrate. One might think that this result is an artifactof Thus at any given point in space, populationsmove the very simple model of populationgrowth. How- fromthe absenceof populationpressure to high pop- ever,it is easy to addmuch realism to themodel with- ulationpressure as the wave passes over thatpoint. out anychange of the basic result.In Appendix A we Figure4 shows the pattemof spreadfor r = .01 and show that a more realistic version of the logistic d = 30. Withthese quiteconservative values, it takes equationactually leads to even more rapid growth less than200 years for the wave frontto pass from of populationpressure. low populationpressure to high populationpressure. AllowingJfr Dispersal. Once, after listening to More realisticmodels thatallow for density-depen- one of us propoundthis argument, a skepticalarchae- dent migration also yield a constant, wave-like ologist replied,"But you've got to fill up all of Asia, advanceof population(Murray 1989), and although first."This understandable intuitive response betrays the rates vary,we believe that the same qualitative a deep misunderstandingof the time scales of expo- conclusionwill hold. nential growth.Suppose that the initial population The Dynamics of Innovation. So far we have of anatomicallymodem humans was only about104 assumed that the carryingcapacity of the environ- andthat the carryingcapacity for hunter-gatherersis ment is fixed (save where it is increasedby fortu- very optimistically 1 person per squarekilometer. itous inventions).However, we know that people Giventhat the land areaof the Old Worldis roughly respondto scarcitycaused by populationpressure by 108 km2 , 0 = 104 /10 8 = 10-4 . Thenusing equation2 intensifying production,for example, by shifting and againassuming r = .01, Eurasiawill be filled to fromless labor-intensiveto morelabor-intensive for- 99 percentof carryingcapacity in about1,400 years. aging, or by innovationsthat increase the efficiency The differencebetween increasing population pres- of subsistence (Boserup 1981). Since innovation sureby a factorof 100 and by a factorof 10,000 is increasescarrying capacity, intuition suggests that it only about500 years! might thereforedelay the onset of populationpres- Moreover,this calculation seriously over-esti- sure. However,as the model in AppendixB shows, mates the amountof time thatwill pass before any this intuition,too, is faulty. segment of an expandingEurasian population will Figure5 shows the resultsof the model in Appen- experiencepopulation pressure because populations dix B. A smallpopulation initially grows rapidly. As will approachcarrying capacity locally long before populationpressure builds, populationgrowth rate theentire continent is filled withpeople. R. A. Fisher slows to a steadystate in which populationpressure (1937) analyzed the following partial differential is constant, and just enough innovationoccurs to equationthat captures the interactionbetween pop- compensate for population growth. For plausible ulationgrowth and dispersalin space: parametervalues the second phase of population 398 AMERICAN ANTIQUITY [Vol. 66, No. 3, 2001

5

4

Population 3 Density 2500 yrs 2000 yrs 1500 yrs 1000 yrs 500 yrs 500 yrs 1000 yrs 1500 yrs 2000 yrs 2500 yrs (person/kM2) 2

250 1 yrs

0 yrs

2500 2000 1500 1000 500 500 1000 1500 2000 2500 Distance (km)

Figure 4. A numerical simulation of Fisher's equation showing that after an initial period, population spreads at a constant rate so that at any point in space population pressure increases to its maximum in less than 500 years for reasonable para- meter values. (Redrawn from Ammerman and Cavalli-Sforza, 1984).

growthsteady state is reachedin less than a thou- Thispicture of the interactionof demographyand sandyears. Interestingly, increasing the intrinsicrate innovationleads to predictionsquite differentfrom of innovationor theinnovation threshold reduces the those of scholarslike Cohen (1977). For example, waitingtime until populationpressure is important. we do not expect to see any systematicevidence of Innovationallows greaterpopulation increases over increasedpopulation pressure immediately prior to the long run,but it does not change the time scales major innovations,an expectationconsistent with on which population pressure occurs. The most the record(Hayden 1995). If people are motivated importantfactor on time scales of a millenniumor to innovatewhenever population pressure rises above greater(if not a centuryor greatergiven realistic an innovationthreshold, and it, in the absenceof suc- startingpopulations) is the rateof intensificationby cessful innovation, populations adjust relatively innovation,not populationgrowth. ouicklv to changes in K by growthor contraction-

Initial SteadyState Population Population GrowthRate GrowthRate

Log Population Size g

0 260 400 600 800 1000 Time (yrs)

Figure5. This plotsthe logarithmof populationsize as a functionof timefor the modeldescribed in AppendixB. Initially, when there is little populationpressure, population grows at a high rate. As the populationgrows, per capita income decreases,and peopleintensify. Eventually the populationgrowth rate approachesa constantvalue at whichthe growthof intensificationbalances growth in population.For reasonable parameters (a = 0.005,r = 0.02,y. = 1,y = 0.1,y. = 0.2, initial populationsize 1 percentof initialcarrying capacity), it takesless than 500 years to shiftfrom the initiallow population pres- suremode of growthto the finalhigh population pressure mode of growth. Richerson et al.] AGRICULTURALORIGINS AND CLIMATE 399 thenevidence of extraordinarystress-for example, sistencesystem to settledvillage agriculturein a fast skeletal evidence of malnutrition-is likely only case. Ten thousandyears in the Holocene was ulti- when rapid environmentaldeterioration exceeds a matelysufficient for the developmentof plant-inten- population'scapacity to respondvia a combination sive gathering technologies or agriculture of downwardpopulation adjustment and innovation.5 everywhereexcept in the coldest, plant-poorenvi- Thus,for parametervalues that seem anywherenear ronments. reasonableto us, populationgrowth on millennial time scales will be limitedby ratesof improvement The Pattern Of Intensification Across Cases in subsistenceefficiency, not by the potentialof pop- Implicates Climate Change ulations to grow, just as Malthusargued. Popula- We have arguedthat Malthusianprocesses lead to tionscan behave in non-Malthusianways only under population pressure much more quickly than extremeassumptions about population dynamics and assumedby such writersas Cohen (1977) and that ratesof intensification,such as the modernworld in the rate of cultural"settling in" and intensification whichthe rateof innovation,but also the rateof pop- is fasterthan Braidwood(1960) imagined,but not ulationgrowth, is very high. fast enough to intensifymore than a small distance Of course,in a time as variableas the Pleistocene, towardagriculture in the highly variableenviron- populationsmay well have spent considerabletime ments of the Pleistocene.Thus, our hypothesisthat both farabove and far below instantaneouscarrying the abrupttransition from glacial to Holocene cli- capacity.If agriculturaltechnologies were quick and mates caused the origin of agriculturerequires that easy to develop, the population-pressureargument Holocenerates of intensificationbe neithertoo slow would lead us to expect Pleistocenepopulations to nor too fast. shift in and out of agricultureand other intensive AgriculturceWas Independently Evolved strategies as they find themselves in subsistence About 10 Times crises due to environmentaldeterioration or in peri- ods of plentydue to amelioration.Most likely,minor The sample of origins is large enough to support intensificationsand de-intensificationswere stan- some generalizationsabout the processes involved. dardoperating procedure in the Pleistocene. How- Table 1 gives a roughtime line for the originof agri- ever, the time needed to progress much toward culture in seven fairly well-understoodcenters of plant-richstrategies was greaterthan the fluctuating domestication,two morecontroversial centers, three climateallowed, especially given CO2-and aridity- areasthat acquired agriculture by diffusion,and two limitedplant production. areas that were without agricultureuntil European conquest.6 The list of independentcenters is com- CulturalEvolution Has the TimeScale Wrong plete as faras currentevidence goes, and while new The timing of the originof agriculturemight possi- centersare not unexpected, it is unlikelythat the pre- bly be explainedentirely by the rateof intensifica- sent list will double.Numerous areas acquired agri- tion by innovation.For example, Braidwood (1960) cultureby diffusion(societies acquiremost of their arguedthat it took some time for humansto acquire technological innovationsby diffusion, not inde- enoughfamiliarity with plantresources to use them pendentinvention), so the threeareas in Table 1 are as a primarysource of calories, and that this "set- but a small sample.The numberof non-arcticareas tling in" process limited the rate at which agricul- withoutagriculture at Europeancontact is small and ture evolved. This proposal may explain the the two listed,western North America and Australia, post-Pleistocenetiming of the developmentof agri- are the largestand best known. culture.However, if we interprethis argumentto be Two lines of evidenceindicate that the sevencen- thatthe settlingin processbegan with the evolution ters of domestication are independent. First, the of behaviorallymodern humans, the time scale is domesticatestaken up in each centerare distinctive wrongagain. There is no evidence thatpeople were and no evidence of domesticatesfrom othercenters makingsignificant progress at all towardagriculture turnsup early in any of the sequences.For example, for 30,000 years, and Braidwood'sexcavations at in the easternNorth American center a sunflower,a Jarmoshow that some 4,000 years was enough to goosefoot, marshelder, an indigenoussquash, and go from un-intensivehunting-and-gathering sub- otherlocal plantswere takeninto cultivationaround 400 AMERICANANTIQUITY [Vol.66, No. 3, 2001

Table 1. Dates Before Present in CalendarYears of Achievementof Plant Intensive Huntingand Gatheringand Agriculturein Different Regions, mainly after Smith (1995).

Region Intensive Foraging Agriculture Center-sof Domestication Near East: Bar-Yosefand Meadow 1995 15,000 11,500 North China:An 1991; Elston et al. 1997 11,600 > 9,000 South China:An 1991 12,000? 8,000 Sub-SaharaiAfrica: Klein 1993 9,000 4,500 SouthcentralAndes: Smith 1995 7,000 5,250 CentralMexico: Smith 1995 7,000 5,750 EasternUnited States: Smith 1995 6,000 5,250

Controversial Centers HighlandNew Guinea:Golson 1977 ? 9,000? Amazonia: Pearsall 1995 13,000? 9,000?

Acquiisitionby Diffusion NorthwesternEurope 12,500 7,000 SouthwesternU.S.: Cordell 1984; Doelle 1999 6000 3,500 Japan:Aikens and Akazawa 1996; Crawford1992 10,500 3,000

Never- Acquired Agricuilturi-e California:Bettinger 2000 4,000 n/a Australia:Hiscock 1994; Smith 1987 3,500 n/a

6000 B.P. Mesoamerican maize subsequently within a thousand years of the beginning of the appearedhere around 2000 B.P., but remained a Holocene,even thoughthe best-documented, clearly minordomesticate until around 1100 B.P., when it agriculturalcomplexes are still considerablylater suddenly crowded out several traditionalcultivars (An 1991; Crawford 1992; Lu 1999). Agriculture (Smith 1989). Second, archaeology suggests that may proveto be as early in northernChina as in the none of the centershad agriculturalneighbors at the NearEast, since theearliest dated sites, which extend time that their initial domesticationswere under- back to 8500 B.P., representadvanced agricultural taken. The two problematiccenters, New Guinea systemsthat must have takensome time to develop. and Lowland South America, present difficult Excavationsin northernChina northof the earliest archaeologicalproblems (Smith 1995). Sites are hard dated agriculturalsites document a technological to findand organic remains are rarely preserved. The change around1 1,600 B.P.,signaling a shift toward New Guineaevidence consists of apparentlyhuman- intensive plant and animal procurementthat may constructedditches that might have been used in have set this process in motion (Elstonet al. 1997). controllingwater for taro cultivation.The absence The exact sequence of events also varies quite of documentedliving sites associatedwith these fea- widely.For example, in theNear East, sedentism pre- tures makes their interpretationquite difficult.The ceded agriculture,at least in the LevantineNatufian lowlandSouth American evidence consists of starch sequence, but in Mesoamericacrops seem to have grainsembedded in potteryfragments and phytoliths, been addedto a hunting-and-gatheringsystem that microscopicsilicious structural constituents of plant was dispersed and long remained rather mobile cell walls. The largesize of some early starchgrains (MacNeish1991:27-29). For example, squash seems and phytolithsconvinces some archaeologiststhat to have been cultivatedaround by 10,000 B.P. in rootcrops were brought under cultivation in theAma- Mesoamerica,some 4,000 years before corn and zon Basin at very early dates. bean domesticationbegan to lead to the origin of a The timingof initiationof agriculturevaries quite fully agriculturalsubsistence system (Smith 1997). widely.The NearEastern Neolithic is the earliestso Some mainlyhunting-and-gathering societies seem far attested. In northern,and possibly southern, to have incorporatedsmall amountsof domesticated China, however, agriculture probably followed plantfoods intotheir subsistence system without this Richerson et al.] AGRICULTURALORIGINS AND CLIMATE 401 leadingto full-scaleagriculture for a verylong time. tion,even if theirless intensiveexploitation was com- PerhapsAmerican domesticates were long used to mon.The broadspectrum of species, includingsmall provide specialized resources or to increase food game and plants,reflected in these cases is not per securitymarginally (Richard Redding, personal com- se evidence of intensification(specialized use of munication) and initially raised human carrying more costly but more productiveresources using capacitiesrelatively little, thus operatingthe com- more labor and dedicatedtechnology), as is some- petitiveratchet quite slowly. According to MacNeish, times argued(Flannery 1971). In most hunter-gath- the path forwardthrough the whole intensification erer systems, marginaldiet cost and diet richness sequencevaried considerably from case to case. (numberof speciesused) are essentially independent (Bettinger 1994:46-47), and prey size is far less A Late Intensificationof Plant Gathering importantin determiningprey cost thaneither mode PrecedesAgriculture or context of capture(Bettinger 1993:51-52; Bet- Inall knowncases, the independentcenters of domes- tingerand Baumhoflf 1983:832; Madsen and Schmitt ticationshow a latesequence of intensificationbegin- 1998). Forall these reasons,quantitative features of ning with a shift from a hunter-gatherersubsistence subsistencetechnology are a better index of Pleis- system based upon low-cost resourcesusing mini- toceneresource intensification than species used. We mal technologicalaids to a system based upon the believethat the dramaticincrease in the quantityand procurementand processing of high-costresources, rangeof small chippedstone and groundstonetools includingsmall game and especially plantseeds or only after 15,000 B.P. signals the beginningof the other labor-intensive plant resources, using an patternof intensificationthat led to agriculture. increasingrange of chippedand groundstone tools Earlyintensification of plantresource use would (Hayden 1995). The reasons for this shift are the havetended to generatethe samecompetitive ratchet subject of much work among archaeologists(Bet- as the laterforms of intensification.Hunter-gather- tinger 2000). The shifts at least accelerate and ers who subsidizehunting with plant-derivedcalo- become widespreadonly in the latestPleistocene or ries can maintainhigher populationdensities, and Holocene. However, a distinct tendency toward thuswill tend to depletebig game to levels thatcan- intensificationis often suggestedfor the UpperPale- not sustainhunting specialists (Winterhalder and Lu olithic more generally.Stiner et al. and commenta- 1997). Upper Paleolithicpeople appearto be fully tors(2000) notethat Upper Paleolithic peoples often modernin theirbehavioral capacities (Klein 1999). madeconsiderable use of small mammalsand birds Importantchanges in subsistence technology did in contrastto earlierpopulations. These species have occur duringthe UpperPaleolithic, for examplethe much lowerbody fat thanlarge animals, and exces- developmentof theatlatl. Nevertheless, modern abil- sive consumptioncauses ammoniabuildup in the ities and the operationof the competitive ratchet body due to limitationson the rateof ureasynthesis droveUpper Paleolithic populations only a relatively ("rabbitstarvation," Cordain et al. 2000). Conse- small distance down the path to the kind of heavy quently,any significant reliance on low-fatsmall ani- relianceon plantresources that in turnset the stage malsimplies corresponding compensation with plant for domestication. calories, and at least a few Upper Paleolithicsites, Braidwood'sreasoning that pioneeringagricul- such as the Ohalo1I settlement on the Sea of Galilee turalistswould have gained their intimatefamiliar- (Kislev et al. 1992), show considerableuse of plant ity with proto-domesticates first as gatherers is materialsin Pleistocenediets. Large-seededannual logical and supportedby the archaeology.Once the species like wild barley were no doubt attractive climateameliorated, the rateof intensificationaccel- resourcesin the Pleistocenewhen presentin abun- eratedimmediately in the case of the Near East. In dance, and would have been used opportunistically other regions changes right at the Pleistocene- duringthe last glacial. If our hypothesisis correct, Holocenetransition were modestto invisible(Straus in the last glacial no one attractivespecies like wild et al. 1996). The full working out of agrariansub- barley would have been consistentlyabundant (or sistence systems took thousandsof years. Indeed, perhapsproductive enough) for a long enough span modernbreeding programs illustrate that we arestill of time in the same location to have been success- workingout the possibilitiesinherent in agricultural fully targetedby an evolving strategyof intensifica- subsistencesystems. 402 AMERICAN ANTIQUITY [Vol. 66, No. 3, 2001

The cases whereHolocene intensification of plant the slowest Holocene rates of intensificationwere gatheringdid not lead directlyto agricultureare as rapidenough to producehighly visible archaeolog- interestingas the cases where it did. The Jomon of ical evidenceon the ten-millenniumtime scale, one- Japan represents one extreme (Imamura 1996). third or less time than Upper Paleolithic peoples Widespreaduse of simple pottery,a markerof well- lived underglacial climates. developedagricultural subsistence in westernAsia, IntensiveTechnologies Tend to Spread was very early in the Jomon,contemporary with the More latest Pleistocene Natufian in the Near East. By One successfuland durable agricultural origin in the 11,000 yrs B.P.,the Jomon people lived in settledvil- last glacial on any sizeable land mass would have lages, dependedsubstantially upon plant foods, and been sufficientto producea highly visible archae- used massive amounts of pottery. However, the ological record, to judge from events in the Jomondomesticated no plantsuntil rather late in the Holocene. Once well-establishedagricultural sys- sequence.Seeds of weedy grassesare found through- tems existed in the Holocene, they expandedat the out, but only in laterphases (afterabout 3000 B.P.) expense of hunting-and-gathering neighbors at do the first unambiguousdomesticates occur, and appreciablerates (Bellwood 1996). Ammermanand these make up only a small portionof the seeds in Cavalli-Sforza(1984) summarizethe movementof archaeologicalcontexts (Crawford 1997). Sophisti- agriculturefrom the Near East to Europe, North cated agriculturecame to Japanwith importedrice Africaand Asia. The spreadinto Europeis best doc- from the mainlandonly about 2,500 B.P. Interest- umented.Agriculture reached the Atlantic seaboard ingly, acorns were a major item of Jomon subsis- about 6000 B.P. or about 4,000 years after its ori- tence. The people of Californiawere anothergroup gins in the Near East. The regularityof the spread, of sedentaryhunter-gatherers that depended heavily andthe degreeto which it was largelya culturaldif- on acorns. However,in Californiathe transitionto fusion process as opposed to a populationdisper- high plantdependence began much laterthan in the sion as well, are mattersof debate. Cavalli-Sforza Jomon(Wohlgemuth 1996). Millingstones for grind- et al. (1994:296-299) argue that demic expansion ing small seeds became importantafter 4500 B.P., by western Asians was an importantprocess with althoughseeds were of relativelyminor importance the front of genes moving at about half the rate of overall.After 2800 B.P. acornsprocessed with mor- agriculture.They imagine that pioneeringagricul- tars and pestles became an importantsubsistence turalpopulations moved into territoriesoccupied by componentand small seeds faded in comparative hunter-gatherers,and intermarriedwith the pre- importance.In the latest period, after 1200 B.P., existing population.The then-mixedpopulation in quantitiesof small seeds were increasinglyadded turn sent agricultural pioneers still deeper into back into the subsistencemix alongsideacorns in a Europe.They also suppose that the rate of spread plant-dominateddiet. Other peoples with a lateonset was fairly steady,though clearly frontiersbetween of intensificationinclude the Australians.The total- hunter-gatherersand agriculturalistsstabilized in ity of cases tells us thatany stage of the intensifica- some places (Denmark,Spain) for relatively pro- tion sequence can be stretchedor compressedby longed periods. Zvelebil (1996) emphasizes the severalthousand years but reversalsare rare(Harris complexityand durability of frontiersbetween farm- 1996; Price and Gebauer1995). Farmingdid give ers and hunter-gatherersand the likelihood that in way to hunting-and-gatheringin the southernand many places the diffusion of both genes and ideas easternGreat Basin of NorthAmerica after a brief about cultivation was a prolonged process of extensionof farminginto the region around 1000 B.P. exchange across a comparativelystable ethnic and (Lindsay1986). A similarreversal occurred in south- economicfrontier. Further archaeological and paleo- ern Sweden between 2400 and 1800 B.P. (Zvelebil genetic investigations will no doubt gradually 1996).Horticultural Polynesian populations returned resolve these debates.Clearly, the spreadprocess is substantiallyto foragingfor a few centurieswhile at least somewhatheterogeneous. populationdensities built up on reachingthe previ- Otherexamples of the diffusionof agricultureare ously uninhabitedarchipelagos of Hawaiiand New relatively well documented. For example, maize Zealand(Kirch 1984). Had intensificationon plant domesticationis datedto about6200 B.P.in Central resourcesbeen possible duringthe last glacial,even Mexico, spreadingto the southwesternU.S. (New Richerson et al.] AGRICULTURALORIGINS AND CLIMATE 403

Table 2. Processes that may Retardthe Rate of CulturalEvolution and Create Local Optimathat Halt Evolution for ProlongedPeriods.

Process Authors (examples) Geography:Eurasia, having the largest land mass, has more local popula- Diamond 1997 tions to exchange innovationsby diffusion, hence the fastest Holocene rate of subsistence intensification.

Minor climate change: The late medieval onset of the Little Ice Age caused Kent 1987; Kleivan 1984 the extinction of the GreenlandNorse colony. Agricultureat marginalalti- tudes in places like the Andes seems to respondto Holocene climatic fluctu- ation.

Preadaptedplants: The MediterraneanOld Worldis unusuallywell endowed Blumler 1992; Blumler and Byrne 1991; with large-seededgrasses susceptible to domesticationpressures. American Diamond 1997; Hillman and Davies 1990 domesticates,especially maize, may outcross too much to respondquickly to selection.

Diseases: Density dependentepidemic diseases may evolve that slow or stop Cavalli-Sforzaet al. 1994; Crosby 1986; the populationgrowth, pending the evolution of resistance,that would other- Gifford-Gonzalez2000; McNeill 1976 wise drive the competitiveratchet. Local diseases that attackforeigners may protectotherwise-vulnerable systems.

New technological complexes evolve slowly: Nutritionaladequacy in plant- Katz et al. 1974 rich diets requiresdiscovering cooking techniques,acquiring balancing domesticates,developing the potentialof animal domesticates,and the like.

New social institutionsevolve slowly: Social institutionsare generally deeply Bettinger 1999; Bettingerand Baumhoff 1982; involved in subsistence but are also liable to be regulatedby norms that North and Thomas 1973; Richerson and Boyd make adaptiveevolution away from currentlocal optima difficult. 1999

Ideology may play a role: The evolution of fads, fashions, and belief systems Weber 1930 may act to drive culturalevolution in non-utilitariandirections that some- times carrythem to new adaptiveslopes.

Mexico) by about4000 B.P. (Matson 1999; Smith intensification.The modernizationof culturecapac- 1995). In this case, the frontierof maize agriculture ities leading up to the Upper Paleolithictransition stabilizedfor a long time, only reachingthe eastern was presumablysuch an event, as were laterinven- U.S. at the comparativelylate date noted above. tions like literacy(Donald 1991;Klein 1999:Ch.7). Maize failed entirely to diffuse westwardinto the We arenot awareof anyproposals for majorchanges Mediterraneanparts of Californiaeven thoughpeo- in the intrinsicrate of culturalevolution coincident ples growingit in the more aridparts of its rangein with the Pleistocene-Holoceneboundary. Students the Southwestused irrigationtechniques that have of the evolution of subsistenceintensification and eventuallyworked in Californiawith modestmodi- social complexityin the Holocenehave suggesteda ficationsto cope with dry-seasonirrigation. As with series of plausibleprocesses that will probablyturn the origin process, the rate of spreadof agriculture out to be at least partof the explanationfor why the exhibitsan interestingdegree of variation. trendto intensificationhas takensuch diverseforms in differentregions (Table2). This list of diversify- Changes in the Cultural ing andrate-limiting processes does not includeany Evolutionary System? thatshould have operated more stringently on Upper A possiblealternative to ourhypothesis would be that Paleolithic,as opposedto Mesolithicand Neolithic, a substantialmodernization of the culturalsystem societies, climate effects aside. Holocene rates of occurredcoincidently at the end of the Pleistocene intensificationdo havethe right time scale to be dras- and that this resulted in a general accelerationof tically affected by millennial- and submillennial- rates of cultural evolution, including subsistence scale variationthat is rapidwith respectto observed 404 AMERICAN ANTIQUITY [Vol. 66, No. 3, 2001 ratesof culturalevolution in the Holocene. and lower averageprecipitation suggest that world If climate variationdid not limit intensification averageagricultural output would fall considerably. duringthe lastglacial to vanishinglyslow ratescom- Currentanthropogenic global warming via green- pared to the Holocene, the failure of intensive house gases might at least temporarilyprevent any systems to evolve during the tens of millennia returnto glacialconditions. However, we understand anatomicallyand culturalmodem humanslived as the feedbacks regulating the climate system too sophisticatedhunter-gatherers before the Holocene poorlyto haveany confidence in suchan effect. Cur- is a considerablemystery. rentincreases in CO2threaten to elevateworld tem- peraturesto levelsthat in pastinterglacials apparently Conclusion triggereda large feedbackeffect producinga rela- The large,rapid change in environmentat the Pleis- tively rapiddecline towardglacial conditions(Petit tocene-Holocenetransition set off the trendof sub- et al. 1999). The Arctic Ocean ice pack is currently sistence intensificationof which modem industrial thinningvery rapidly (Kerr 1999). A dark,open Arc- innovations are just the latest examples. If our tic Ocean would dramaticallyincrease the summer hypothesisis correct,the reductionin climate vari- heatincome at high northern latitudes and have large, ability,increase in CO2content of the atmosphere, difficult-to-guessimpacts on the Earth'sclimate sys- and increasesin rainfallrather abruptly changed the tem. No one can yet estimatethe riskswe aretaking earthfrom a regimewhere agriculture was impossi- of a rapidreturn to colder,drier, more variable envi- ble everywhereto one whereit was possiblein many ronmentwith less CO2, norevaluate exactly the threat places. Since groups that use efficient, plant-rich such conditionsimply for the continuationof agri- subsistence systems will normally out-compete culturalproduction. Nevertheless, the intrinsicinsta- groups that make less efficient use of land, the bility of the Pleistocene climate system, and the Holocenehas been characterizedby a persistent,but degreeto whichagriculture is likely dependentupon regionallyhighly variable,tendency toward subsis- the Holocene stable period, should give one pause tence intensification.The diversity of trajectories (Broecker1997). taken by the various regional human sub-popula- tionssince = 11,600 B.P. are natural experiments that Acknowledgments.We thank Joe Andrew, Ofer Bar-Yosef, Richard Redding, Bruce Winterhalder,and three anonymous will help us elucidate the factors that control the referees for unusuallyconstructive criticism of the manuscript. tempo of culturalevolution and that generatehis- Thanks to Scott Elias for insights pertaining to Pleistocene torical contingency against the steady, convergent seasonalityand to PeterDitlevsen for providingFigure 2. Peter adaptivepressure toward ever moreintense produc- Lindert's invitation to give a seminar led to the first draft of tion systems.A long list of processes(Table 2) inter- this paper.Thanks to Francisco Gil-White for assistance with the Spanish abstract. actedto regulatethe nearlyunidirectional trajectory of subsistence intensification,population growth, Appendix A: and institutionalchange that the world's societies More Realistic Population Dynamics have followed in the Holocene. Notably, even the slowest evolving regions generatedquite apprecia- The logistic equationassumes that an incrementto ble and archaeologically visible intensification, populationdensity has the sameeffect on population demandingsome explanationfor why similar tra- pressureat low densities as at high densities. We jectories are absentin the Pleistocene. know thatthis assumptionis not correctin all cases. Thosewho arefamiliar with the Pleistoceneoften For example, hunterspursuing herd animals may remarkthat the Holocene is just the "presentinter- generate much populationpressure at low human glacial."The returnof climatevariation on the scale populationdensities because killing only a small that characterizedthe last glacial is quite likely if fractionof the herdmakes the manysurvivors diffi- currentideas aboutthe Milankovichdriving forces cult to hunt.On the otherhand, subsistence farmers of the Pleistoceneare correct. Sustaining agriculture spreadinginto a uniformfertile plain may feel little underconditions of much higher amplitude,high- populationpressure until all farmlandis occupied. frequencyenvironmental variation than farmers cur- If returnsto additionallabor on shrinkingfarms then rentlycope with would be a considerabletechnical dropsteeply, most populationpressure will be felt at challenge.At thevery best, lower C02 concentrations densitiesnear K. To allowfor sucheffects, ecologists Richerson et al.] AGRICULTURALORIGINS AND CLIMATE 405 often utilize a generalizedlogistic equation dN = pN(y - YS) (B2) dt dNN = rNIl- - where is the per capitaincome necessaryfor sub- dt ( (K)) ((Al) 1 y, sistence. If per capitaincomes are above this value, Populationpressure is now givenby the term(N/K)0 populationincreases; if percapita income falls below If 0 > 1, populationpressure does not increaseuntil y,, populationshrinks. If I is fixed, this equationis densitiesapproach carrying capacity, as is usuallythe anothergeneralization of the logistic equation.In an case for specieslike humansthat have flexible behav- initially empty environment,population initially ior andconsiderable mobility, and thus can mitigate grows at a ratep(y,,, -y), butthen slows andreaches the effects of increasing population density over an equilibriumpopulation size some rangeof densities. It seems intuitivethat this - would increasethe lengthof time necessaryto reach I(Y111Ys ) (B3) a given level of populationpressure. However, this Ys intuitionis wrong. The generalizedlogistic can be To allow for intensificationwe assumethat peo- used to derivea differentialequation for XT= (N/K)0 ple innovatewhenever their per capitaincome falls dir 0 (N0 dN below a thresholdvalue yi. Thus dt N1K) dt dI - = aI(y1I y) (B4) dt I( f 2)) (A2) N When per capita income is less than the threshold value Yi, people innovate, increasing the carrying - =i0(I ir) capacityand thereforedecreasing population pres- Thus, the differentialequation for populationpres- sure. When per capita income is greaterthan the sureis alwaysthe ordinary logistic equation in which threshold,they "de-innnovate."This may seem odd K = 1 and r is multipliedby 0. This meansthat when atfirst, but such abandonment of moreefficient tech- 0 > 1, it takes less time to reach a given amountof nology has been observedoccasionally. The maxi- populationpressure than would be the case if 0 = 1. mumrate at which innovationcan occuris governed Reducedpopulation pressure at low densities leads by the parametera. to more rapidinitial populationgrowth. Population If a small pioneer populationenters an empty growthis close to exponentiallonger and this more habitat,it experiencestwo distinctphases of expan- than compensatesfor the fact that higher densities sion (Figure5). Initially,per capita income is near have to be reachedto achievethe same level of pop- themaximum, and population grows at the maximum ulationpressure. rate. As population density increases, per capita income dropsbelow yi, andthe populationbegins to Appendix B: The Dynamics of Innovation innovate,eventually reaching a steady state value Considera populationof size N in which the per capitaincome of the populationis given by: PYS+ ayi (B5) p+a y=I+N (B1) The steady state per capita income is above the wherey,, is the maximumper capitaincome, and I minimumfor subsistencebut below the thresholdat is a variablethat represents the productivityof sub- which people experience populationpressure and sistencetechnology. Thus per capita income declines begin to innovate.At this steady state population as populationsize increases,but for a given popula- growthcontinues at a constantrate, tion size, greater productivity raises per capita income.As in the previousmodels, we assumethat a(y1-y) (B6) as populationpressure, now measuredas falling per p+a capita income, increases, population growth that is proportionalto the rate of growthof subsis- decreases.In particular,assume: tence efficiency. 406 AMERICAN ANTIQUITY [Vol. 66, No. 3, 2001

References Cited 1996 The Origins and Spread of Agriculturein the Indo- PacificRegion: Gradualism and Diffusion or Revolution Aikens, C. M., andT. Akazawa and Colonization.In TheOrigins and SpreadofAgriculture and 1996 The Pleistocene-HoloceneTransitionin JapanandAdja- Pastoralismin Eurasia,edited by D. R. Harris,pp. 465-498. cent NortheastAsia: Climate and Biotic Change, Broad- SmithsonianInstitution Press, Washington, D.C. SpectrumDiet, Pottery,and Sedentism. In Humansat the Bennett,K. D., S. G. Haberle,and S. H. Lumley End of the Ice Age, edited by L. G. Straus,B. V. Eriksen,J. 2000 TheLastGlacial-HoloceneTransitionin Southern Chile. M. Erlandson,and D. R.Yesner, pp.215-227. Plenum Press, Science New York. 290:325-328. Belovsky,G. E. Allen, J. R., U. Brandt,A. Brauer,H.-W. Hubberten, B. Huntley, 1987 Hunter-GathererForaging: A Linear Programming J. Keller,M. Kraml,A. Mackensen,J. Mingram,J. F.Negen- Approach.Journal ofAnthropologicalArchaeology 6:29-76. dank,N. R. Nowaczyk,H. Oberhansli,W. A. Watts,S. Wulf, Bettinger,R. L. and B. Zolitschka 1991 Hunter-Gatherers:Archaeological and Evolutionary 1999 RapidEnvironmental Changes in SouthernEurope Dur- Theory.Plenum, New York. ing the Last GlacialPeriod. Nature 400:740-743. 1993 Doing GreatBasinArchaeologyRecently: CopingWith Allen, J. R. M., W. A. Watts,and B. Huntley Variability.Journal of ArchaeologicalResearch 1:43-66. 2000 WeichselianPalynostratigraphy, Palaeovegetation and 1994 How, When, and Why Numic Spread. In Across the Palaeoenvironment:The RecordFrom Logo Grandedi Mon- West:Human Population Movement and the Expansionof ticcio, Southern Italy. Quaternary International the Numa, edited by D. B. Madsen 73/74:91-110. and D. R. Rhode, pp. 44-55. Universityof Utah Press, Salt Lake City. Ammerman,A. J., and L. L. Cavalli-Sforza 1999 From Travelerto Processor:Regional Trajectoriesof 1984 The Neolithic Transitionand the Genetics of Popula- Hunter-GathererSedentism in the Inyo-MonoRegion, Cal- tions in Europe.Princeton University Press, Princeton. ifornia.In SettlementPattern An, Z. Studiesin theAmericas: Fifty YearsSince Viru,edited by B. R. Billman and G. M. Fein- 1991 RadiocarbonDating and the PrehistoricArchaeology man,pp. 39-55. SmithsonianInstitution Press, Washington, of China.WorldArchaeology 23:193-200. D.C. Bar-Yosef,0. 2000 HoloceneHunter-Gatherers. InArchaeology at theMil- 1998 The NatufianCulture in the Levant,Threshold to the lennium:A Sourcebook,edited by G. M. Feinmanand T. D. Origins of Agriculture. Evolutionary Anthropology Price, pp. 137-195. Plenum,New York. 6:159-177. Bettinger,R. L., and M. A. Baumhoff Bar-Yosef,O., and R. H. Meadow 1982 The Numic Spread:Great Basin Culturesin Competi- 1995 The Origins of Agriculturein the Near East. In Last tion. AmericanAntiquity 47:485-503. Hunters,First Farmers: New Perspectiveson thePrehistoric 1983 ReturnRates and Intensityof ResourceUse in Numic Transitionto Agriculture, edited by T. D. Price and B. and PrenumicAdaptive Strategies. American Antiquity 48: Gebauer,pp. 39-94. School of American ResearchPress, 830-834. SantaFe. Blumler,M. A. Bar-Yosef,O., and F. R. Valla 1992 Seed Weightand Environmentin Mediterranean-type 1991 TheNatufian Culture in theLevant. International Mono- Grasslandsin Californiaand Israel. Ph.D. Thesis, Univer- graphsin Prehistory,Archaeological Series 1, Ann Arbor. sity of California,Berkeley. UMI, Ann Arbor. Barnola,J. M., D. Raynaud,Y. S. Korotkevich,and C. Loris Blumler,M. A., and R. Byrne (A. Belfer-Cohen,R. M. Bird,V. 1987 Vostok Ice Core Provides 160,000-Year Record of L. Bohrer,B. F. Byrd, R. C. Dunnell,G. Hillman,A. T. M. AtmosphericCO2. Nature329:408-414. Moore, D. I. Olszewski, R. W. Baumhoff,M. A. Redding, and T. J. Riley, Commentators) 1963 EcologicalDeterminants of AboriginalCalifornia Pop- 1991 The EcologicalGenetics of Domesticationand the Ori- ulations.University of CaliforniaPublications in American gins of Agriculture.Current Anthropology 32:23-53 Archaeologyand Ethnology49(2): 155-236. Bond, G., W. Broecker,S. Johnsen,J. McManus,L. J. Bazzaz, F., andW. Sombroek Labeyrie, Jouzel, and G. Bonani 1996 Global ClimateChange and AgriculturalProduction. 1993 CorrelationsBetween Climate Records from North Wiley,Chichester, England. Atlantic Sediments and Beerling,D. J. Greenland Ice. Nature 365(6442):143-147. 1999 New Estimatesof CarbonTransfer to Terrestrial Ecosys- Boserup,E. temsBetween the LastGlacial Maximum and the Holocene. 1981 Populationand Technological Change: A TerraNova 11:162-167. Studyof Long- TermTrends. University of ChicagoPress, Chicago. Beerling,D. J., W. G. Chaloner,B. Huntley,J. A. andM. Peason, Boyd, R., and P. J. Richerson J. Tooley 1985 Cultureand the EvolutionaryProcess. University of 1993 StomatalDensity Responds to theGlacial Cycle of Envi- ChicagoPress, Chicago. ronmentalChange. Proceedings of theRoyal Societyof Lon- Bradley,R. S. don B 251:133-138. 1999 Paleoclimatology:Reconstructing Climates the Beerling,D. J., and F. I. Woodward of Qua- ternary,Second Edition.Academic Press,San Diego. 1993 EcophysiologicalResponses of Plants to the Global Braidwood,R. J. EnvironmentalChange Since the Last Glacial Maximum. 1960 The Agricultural Revolution. American NewvPhytologist 125:641-648. Scientific 203:130-148. Behling, H., H. W. Arz, J. Patzold,and G. Wefer Braidwood,R. J., and B. Howe 2000 Late QuaternaryVegetation and ClimateDynamics in 1960 PrehistoricInvestigations in IraqiKurdistan. Studies in NortheasternBrazil, Inferences from MarineCore GeoB AncientOriental Civilization 31. Universityof ChicagoOri- 3104-1. QuaternaryScience Reviews 19:981- 994. ental Institute,Chicago. Bellwood, P. Richerson et al.] AGRICULTURALORIGINS AND CLIMATE 407

Braidwood,L., R. Braidwood,B. Howe, C. Reed, and P. J. Wat- Cronin,T. M. son (editors) 1999 Principles of Paleoclimnatology.Columbia University 1983 PrehistoricArchaeologyAlongthe Zagios Flanks.Stud- Press, New York. ies in Ancient Oriental Civilization 105. University of Crosby,A. W. ChicagoOriental Institute, Chicago. 1986 Ecological Imperialism:The Biological Expansionof Broecker,W. S. Europe,900-1900. CambridgeUniversity Press, Cambridge. 1995 TheGlacial WorldAccording to Wcally.Lamont-Doherty Darwin,C. EarthObservatory of ColumbiaUniversity, Palisades, NY. 1902 [1874] The Descent of Mcanand Selection and Selec- 1996 GlacialClimate in theTropics.Science 272:1902-1903. tion in Relationto Sex.American Home Library,New York. 1997 ThermohalineCirculation, theAchilles Heel of OurCli- Dansgaard,W., S. J. Johnsen,H. B. Clausen,D. Dahl-Jensen,N. mate System:Will Man-MadeCO2 Upset the CurrentBal- S. Gundestrup,C. U. Hammer,C. S. Hvidberg,J. P. Stef- ance? Science 178:1582-1588. fensen, A. E. Sveinbjonsdottir,J. Jousel, and G. Bond Broecker,W. S., and G. H. Denton 1993 Evidence for GeneralInstability of Past Climatefrom 1989 The Role of Ocean-AtmosphereReorganizations in a 250-kyrIce-core Record. Ncature 364:218-220. Glacial Cycles. Geochimica et Cosmochimica Acta deMenocal,P. B., and J. Bloemendal 53:2465-2501. 1995 Plio-Pleistocene Climatic Variabilityin Subtropical Cavalli-Sforza,L. L., P. Menozzi, andA. Piazza Africa and the Paleoenvironmentof HominidEvolution: A 1994 The Historyand Geogrcaphyof HumnanGenes. Prince- CombinedData-Model Approach. In Paleoclimate and Evo- ton UniversityPress, Princeton. lution WithEmphasis on HumnanOtigins, edited by E. S. Childe,V. G. Vrba, G. H. Denton, T. C. Partridge,L. H. Burckle, pp. 1951 McanMcakes Himself Watts,London. 262-288. Yale UniversityPress, New Haven. Clapperton,C. Diamond,J. 2000 InterhemisphericSynchroneity of MarineIsotope Stage 1997 Guns, Ger-ms,and Steel. Norton,New York. 2 GlacierFluctuations Along theAmerican Cordilleras Tran- Ditlevsen,P. D., H. Svensmark,and S. Johnsen sect. Jolurnalof Quater-naryScience 15:435- 468. 1996 ContrastingAtmospheric and Climate Dynamics of the Clark,P. U., R. B. Alley, and D. Pollard Last-Glacialand Holocene Periods.Ncatute 379:810-812. 1999 NorthernHemisphere Ice-Sheet Influenceson Global Doelle, W. H. ClimateChange. Science 286:1104-1111. 1999 Early Maize in the GreaterSouthwest. Atchaeology Coale, A. J. Southwest13:1-9. 1986 The Decline of Fertilityin EuropeSince the Eighteenth Donald,M. Centuryas a Chapterin HumanDemographic History. In 1991 Originsof the ModetnMind. Harvard University Press, TheDecline in Fertilityin Eluope, editedby A. J. Coale and Cambridge,MA. S. C. Watkins,pp. 1-30. PrincetonUniversity Press, Prince- Dorale,J. A., R. L. Edwards,E. Ito, and L. A. Gonzalez ton. 1998 Climate and VegetationHistory of the Midcontinent Cohen, M. N. from75 to 25 ka:A SpeleothemRecord from Crevice Cave, 1977 TheFood Crisis in Prehistoty:Overpopulation and the Missouri,USA. Science 282:1871-1874. Or-iginsofAgriculture. Yale UniversityPress, New Haven. Downing,T. E., A. J. Olsthoorn,and R. S. J. Tol Cohen,M. N., and G. J. Armelagos 1999 Climate,Change and Risk.Routledge, London 1984 Paleopathologyat the Origins of Agriculture.Acade- Durham,W. H. mic Press,Orlando. 1991 Coevolution:Genes, Culture,and Human Diversity. Coope, G. R. StanfordUniversity Press, Stanford. 1987 The Responseof LateQuaternary Insect Communities Eakin,H. to SuddenClimatic Changes. In Organizationof Communi- 2000 SmallholderMaize Productionand Climatic Risk: A ties Past and Present,edited by J. H. R. Gee andP. S. Giller, Case Study from Mexico. ClimaticChange 45:19-36. pp. 421-438. Oxford,Blackwell. Elias, S. A. Cordain,L., J. B. Miller,S. B. Eaton,N. Mann,S. H. A. Holt,and 1999 Mid-WisconsinSeasonal Temperatures Reconstructed J. D. Speth from Fossil Beetle Assemblagesin EasternNorth America: 2000 Plant-AnimalSubsistence Ratios and Macronutrient Comparisonswith OtherProxy Recordsfrom the Northern Energy Estimationsin WorldwideHunter-Gatherer Diets. Hemisphere.Journal of QuaternaryScience 14:255-262. Amer-icanJournal of ClinicalNutrition 71:682- 692. Elston, R. G., C. Xu, D. B. Madsen, K. Zhong, R. L. Bettinger, Cordell,L. S. J. Li, P. J. Brantingham,H. Wang,and J. Yu 1984 Prehistoryof theSouthwest. Academic Press, San Fran- 1997 New Dates For the NorthChina Mesolithic. Antiquity cisco. 71:985-993. Cowling, S. A., and M. T. Sykes Fiedel, S. J. 1999 PhysiologicalSignificance of Low AtmosphericCO2 1999 Older Than We Thought: Implicationsof Corrected for Plant-ClimateInteractions. Quaternary Research 52: Dates for Paleoindians.American Antiquity 64:95-115. 237-242. Fisher,R. A. Crawford,G. W. 1937 The Waveof Advanceof AdvantageousGenes. Annals 1992 PrehistoricPlant Domestication in EastAsia.In The Ori- of Eugenics(London) 7:355-369. gins of Agriculture,edited by C. W. Cowan and P. J. Wat- Flannery,K. V. son, pp. 7-38. SmithsonianInstitution Press, Washington, 1971 The Originsand Ecological Effects of EarlyDomesti- D.C. cation in Iranand the Near East. In TheDomestication and 1997 Anthropogenesisin PrehistoricNortheastern Japan. In Exploitationof Plants and Animals,edited by S. Struever, People, Plants, and Landscapes:Studies in Paleoethnob- pp. 50-79. AmericanMuseum Sourcebooksin Anthropol- otany,edited by K. J. Gremillion,pp. 86-103. Universityof ogy. NaturalHistory Press, Garden City, New York. AlabamaPress, Tuscaloosa. 1973 The Originsof Agriculture.Annual Reviewv of Anthro- 408 AMERICANANTIQUITY [Vol. 66, No. 3, 2001

pology 2:271-310. Hiscock, P. 1986 Guild Naquitz: Archaic Foragingand EarlyAgr-icul- 1994 TechnologicalResponse to Risk in HoloceneAustralia. ture in Ocaxaca,Mexico. Academic Press, Orlando. Journalof WorldPrehistory 8:267-292. Frogley,M. R., Tzedakis,P. C., andT. H. E. Heaton Imamura,K. 1999 ClimateVariabilityin NorthwesternGreece During the 1996 Jomon and Yoyoi: The Transitionto Agriculture in Last Interglacial.Science 285:886-1889. JapanesePrehistory. In The Or-iginsand Spr-eadofAgr-icul- Gifford-Gonzales,D. tureand Pastoralismin Eurasia,edited by D. R. Harris,pp. 2000 Animal Disease Challengesto the Emergenceof Pas- 442-464. UniversityCollege LondonPress, London. toralism in Sub-SaharanAfrica. Afr-icanAr-chaeological Johnsen,S. J., H. B. Clausen,W. Dansgaard,N. S. Gundestrup, Review,in press. C. U. Hammer,U. Andersen,K. K. Andersen,C. S. Hvid- Golson, J. berg,D. Dahl-Jensen,J. P. Steffensen,H. Shoji,A. E. Svein- 1977 No Room at the Top:Agricultural Intensification in the bjornsdottir,J. White,J. Jouzel, and D. Fisher New Guinea Highlands.In Sunda and Sahul: Prehistoric 1997 The '80 RecordAlong the GreenlandIce Core Project Studiesin SoutheastAsia, Melanesia,and Australia,edited Deep Ice Coreand the Problemof PossibleEemian Climatic by J. Allen, J. Golson, and R. Jones, pp. 601-638. Acade- Instability. Jour-nal of Geophysical Resear-ch C mic Press,London. 102:26,397-26,410. Gommes,R. Kane,S., and G. Yohe 1999 ProductionVariability and Losses. UnitedNations Food 2000 SocietalAdaptation to ClimateVariability and Change: andAgriculture Organization, www.fao.org/sd/eidirect/agro- An Introduction.Climatic Change 45:14. clim/losses.htm. Katz,S. H., M. L. Hediger,and L. A. Valleroy Goring-Morris,N., andA. Belfer-Cohen 1974 TraditionalMaize ProcessingTechniques in the New 1998 The Articulationof CulturalProcesses and Late Qua- World.Science 184:765- 773. ternary EnvironmentalChange In Cisjordan. Palkorient Kent,J. D. 23:71-93. 1987 PeriodicAridity and PrehispanicTiticaca Basin Settle- Grafenstein,U. von, H. Erlenkeuser,A. Brauer,J. Jouzel, and S. ment Patterns.In Ar-idLand Use Str-ategiesand RiskMan- J. Johnsen agement in the Andes, edited by D. L. Browman, pp. 1999 A Mid-EuropeanDecadal Isotope-Climate Record from 297-314. Westview,Boulder. 15,500 to 5000 YearsB.P. Science 284:1654-1657. Kerr,R. A. GRIP(Greenland Ice-core Project Members) 1999 Will theArctic Ocean Lose All Its Ice?Natute 286:128. 1993 ClimateInstability During the Last InterglacialPeriod Kirch,P. V. RecordedIn the GRIPIce Core. Nature364:203-207. 1984 The Evolution of PolynesianiChiefdoms. Cambridge Grove,J. M. UniversityPress, Cambridge. 1988 TheLittle Ice Age. Methuen,London. Kislev,M. E., D. Nadel, and I. Carmi Harris,D. R. (editor) 1992 Epipaleolithic(19,000 BP) Cereal and Fruit Diet at 1996 The Originisand Spiead of Agricultuteand Pastoral- Ohalo II,Sea of Galilee, Israel.Review of Palaoebotanyand ism in Eurasia.University College LondonPress, London. Palynology73:161-166. Harris,D. R. Klein, R. G. 1977 AlternativePathways Toward Agriculture. In Origi1sof 1993 Hunter-Gatherersand Farmersin Africa. In People of Agriculture,edited by C. A. Reed, pp. 179-243. Mouton, the StoneAge: Hunter-Gatherersand Ear-lyFar-mers, edited The Hague. by G. Burenhult,pp. 3947, 50-55. Harper,San Francisco. Hayden,B. 1999 The HumaniCareer: HumaniBiological and Cultural 1981 Researchand Developmentin the Stone Age: Techno- Or-igins.University of Chicago Press, Chicago. logical Transitions Among Hunter-Gatherers.Curri-ent Kleivan,1. Anthropology22:519-548. 1984 History of Norse Greenland.In Handbookof North 1995 A New Overviewof Domestication.In Last Hunters, AmericaniIndianis 5: Arctic, edited by D. Damas, pp. First Farmers:New Perspectiveson) the PiehistoricTransi- 549-555. SmithsonianInstitution, Washington, DC. tion1to Agriculture,edited by T. D. Price and B. Gebauer, Koenig,W. D., R. L. Mumme,W. J. Car-men,and M. T. Stanback pp.273-299. School of AmericanResearch Press, Santa Fe. 1994 Acorn Productionby Oaks in CentralCoastal Califor- Hendy,I. L., and J. P. Kennett nia:Variation Within and AmongYears. Ecology 75:99-109. 1999 Dansgaard-OeschgerCycles andthe California Current Lamb,H. H. System: PlanktonicForaminiferal Response to Rapid Cli- 1977 ClimaticHistoty and the Future.Princeton University mate Change in SantaBarbara Basin, Ocean Drilling Pro- Press, Princeton. gramHole 893A. Paleoceanography15:30-42. Lindsay,L. W. Henry,D. 0. 1986 FremontFragmentation. In Anthr-opology of the Desert 1989 FromForaging to Agriculture:The LevantAt the End West:Essays in Honor of Jesse D. Jenninigs,edited by C. J. of the Ice Age. Universityof PennsylvaniaPress, Philadel- Condie and D. D. Fowler, pp. 229-252. Anthropological phia. PapersNo. I 10. Universityof Utah, Salt LakeCity. Heusser,L. E. Lu,T. L. D. 1995 PollenStratigraphy and Paleoecologic Interpretation of 1999 TheTrcansitionfivom Foracging to Fcarmingacnd the Ori- the 160-K.Y.Record from the Santa BarbaraBasin, Hole gin ofAgr-icultut-ein Chinca.BAR InternationalSeries #774. 893A. Proceedingsof the Ocean Drilling Program,Scien- BritishArchaeological Reports, Oxford. tific Results 146:265-277. MacNeish,R. S. Hillman,G. C., and M. S. Davies 1991 The Or-iginsofAgr-icultut-e acnd Settled Life. University 1990 MeasuredDomestication Rates in Wild Wheats and of OklahomaPress, Norman. BarleyUnder Primitive Cultivation, and Their Archaeolog- Madsen,D. B. ical Implications.Jour-nal of Wor-ldPrehistory 4:157-222. 1994 Mesa Verde and Sleeping Ute Mountain:The Geo- Richerson et al.] AGRICULTURALORIGINS AND CLIMATE 409

graphicaland Chronological Dimensions of Numic Expan- Poli, M. S., R. C. Thunell,and D. Rio sion, In Acrossthe West:Human Population Movlement and 2000 Millennial-ScaleChanges in NorthAtlantic Deep Water the Expansionof the Numa,edited by D. B. Madsenand D. CirculationDuring Marine Isotope Stages 11 and 12: Link- Rhode,pp. 24-31. Universityof UtahPress, Salt LakeCity. age to AntarcticClimate. Geology 28:807- 810. Madsen,D. B., and D. N. Schmitt Price,T. D., and B. Gebauer 1998 Mass Collectingand the Diet BreadthModel: A Great 1995 List Hunters,First Farmers:Newv Perspectives on) the Basin Example. Joulrnal of Archaeological Science PrehistoricTranisitioni to Agriculture.School of American 25:445-455. ResearchPress, SantaFe. Matson,R. G. Reed, C. A. 1999 The Spreadof Maize to the ColoradoPlateau. Arthae- 1977 The Originsof Agriculture.Mouton, The Hague. ology Southwvest13: 10- 1. Reilly, J., and D. Schimmelpfennig Mayewski,P. A., M. L. D., W. S, M. S. Twickler,M. C. Morri- 2000 Irreversability,Uncertainty, and Learning:Portraits of son, R. B. Alley, P. Bloomfield,and K. Taylor Adaptationto Long-TermClimate Change. Climatic Change 1993 The AtmosphereDuring the YoungerDryas. Science 45:253-278. 261:195-200. Richards,B. W., L. A. Owen, and E. J. Rhodes McManus,J. F., D. W. Oppo, and J. L. Cullen 2000 Timingof LateQuaternary Glaciations in the Himalayas 1999 A 0.5-Million-YearRecord of Millennial-ScaleClimate of Northern Pakistan. Journ1al of Quaternaty Science Variabilityin the NorthAtlantic. Science 283:971-975. 15:283-297. McNeill, W. H. Rindos, D. 1976 Plagues and Peoples.Anchor, Garden City, NY. 1984 The Or-igin1sof Agriculture:An EvolutionatyPerspec- Milton, K. tive. AcademicPress, London. 2000 Hunter-GathererDiets-A DifferentPerspective. Amer- Rosenzweig,C., and D. Hillel icaniJournal of CliniicalNutr-itioni 71 (3):665-667. 1998 ClimateChange and the Global Harvest. OxfordUni- Miracle,P. T., and C. J. O'Brien versityPress, New York. 1998 Seasonalityof Resource Use and Site Occupationat Sage, R. F. Badanj, Bosnia-Herzegovinia: Subsistence Stress in an 1995 Was Low AtmosphericCO2 during the Pleistocene a Increasingly Seasonal Environment'?In Seasonality and Limiting Factor for the Origin of Agriculture'?Global Sedentism:A r chaeological Perspectives from Old and Newv ChangeBiology 1:93-106. WorldSites, edited by T. R. Rocek and 0. Bar-Yosef,pp. Sahlins,M. D., and E. R. Service 41-74. Peabody Museum of Archaeologyand Ethnology 1960 Evolutionand Culture.University of MichiganPress, Bulletin6. HarvardUniversity, Cambridge. Ann Arbor. Moore,A. M. T., and G. C. Hillman Schneider,S. H., W. E. Easterling,and L. 0. Mearns 1992 The Pleistoceneto the HoloceneTransition and Human 2000 Adaptation:Sensitivity to NaturalVariability, Agent Economy in SouthwestAsia: The Impactof the Younger Assumptions and Dynamic Climate Changes. Climatic Dryas.American Antiquity 57:482-495. Chacnge45:203-221. Murray,J. D. Schulz, H., E. von Rad, and H. Erlenkeuser 1989 MathematicalBiology. Springer-Verlag,Berlin. 1998 CorrelationBetween ArabianSea and GreenlandCli- Newnham,R. M., and D. J. Lowe mate Oscillations of the Past 110,000 Years. Nature 2000 Fine-ResolutionPollen Record of Late-GlacialClimate 393:54-57. Reversalfrom New Zealand.Geology 28:759-762. Shi, N., L. M. Dupont,H-J. Beug, and R. Schneider North,D. C., and R. P.Thomas 2000 Correlationbetween Vegetation in SouthwesternAfrica 1973 The Rise of the WesternWorld: A New EconomicHis- and Oceanic Upwelling in the Past 21,000 Years.Quater- tory. CambridgeUniversity Press, Cambridge. nary Research54:72-80. Partridge,T. C., G. C. Bond, C. J. H. Hartnady,P. B. deMenocal, Sowers,T., and M. Bender and W. F. Ruddiman 1995 ClimateRecords Covering the Last Deglaciation.Sci- 1995 ClimaticEffects of Late Neogene Tectonismand Vul- ence 269:210-214. canism. In Paleoclimateand EvolutionWith Emphasis on Smith,B. D. Human1 Origin1s, edited by E. S. Vrba, G. H. Denton, T. C. 1989 Originsof Agriculturein EasternNorth America. Sci- Partridge,L. H. Burckle,pp. 8-23. Yale UniversityPress, ence 246:1566- 1571. New Haven. 1995 The Emergence of Agriculture. Scientific American Pearsall,D. M. Library,New York. 1995 Domesticationand Agriculture in the New WorldTrop- 1997 InitialDomestication of Curcurbitapepo in the Amer- ics. In L,astHunters, First Farmers: New Perspectiveson the icas 10,000Years Ago. Science 276:932-934. PrehistoricTransition to Agricultume,edited by T. D. Price. Smith,M. A. andB. Gebauer,pp. 157-192. School of AmericanResearch 1987 Antiquityof Seedgrindingin AridAustralia. Archaeol- Press, SantaFe. ogy in Oceaniai21:29-39. Petersen,L. C., G. H. Haug, K. A. Haugen,and U. Rohl Steig, E. J., E.J. Brook,J. W. C. White,C. M. Sucher,M. L. Ben- 2000 RapidChanges in the HydrologicCycle of the Tropi- der, S. J. Lehman,D. L. Morse, E. D. Waddington,and G. calAtlantic During the Last Glacial. Science 290:1947-1951. D. Glow Petit,J. R., J. Jouzel, D. Raynaud,N. I. Barkov,J.-M. Barnola,J. 1998 SynchronousClimate Changes in Antarcticaand the Basile, M. Bender,J. Cappellaz,M. Davis, G. Delaygue,M. NorthAtlantic. Science 282:92-95. Delmotte,V. M. Kotlyakov,M. Legrand,V. Y. Lipenkov,C. Stiner,M. C., N. D. Munro,and T. A. Surovell (G. Bar-Oz,T. Lorius,L. P6pin,C. Ritz, E. Saltzman,and M. Stievenard Dayan,N. F. Bicho, A. Deitti, J-P.Brugal, E. Carbonell,K. 1999 Climateand AtmosphericHistory of the Past420,000 V. Flannery,S. Newton,A. Pike-Tay,commentators) Years From the Vostok Ice Core, Antarctica. Nature 2000 TheTortoiseand the Hare:Small-Game Use, the Broad- 399:429-436. SpectrumRevolution, and PalolithicDemography. Clri-ent 410 AMERICANANTIQUITY [Vol. 66, No. 3, 2001

Anthropology41:39-73. either labor or technical intensification and since increases in Straus,L. G., B. V. Eriksen,J. M. Erlandson,and D. R. Yesner efficiency usually also lead to population growth, we use the 1996 Hlumansat the End of the Ice Age: TheArchaeology of term "intensification"loosely for the interlinkedprocesses of the Pleistocen1e-Holocen1e Tr-cinsition1. Plenum, New York. labor and technical intensification and population growth. Stuiver,M., P. J. Reimer,E. Bard,J. W. K. Beck, G. S. Burr, A. We define "agriculture"as dependence upon domesticated Hughen,B. Kromer,G. McCormack,J. V. D. Plicht,and M. Spurk crops and animals for subsistence. We mark the origin of 1998 INTCAL98Radiocarbon Age Calibration,24,000 - 0 agricultureas the first horizon in which plant remains having cal BP. Radiocarbon40:1041-1083. anatomical markersof domestication are found, or are likely UnitedNations Food andAgriculture Organization on other grounds to be found in the future. Fully agricultural 2000 GlobalInformation and Early Warning System for Food subsistence systems in the sense of a dominance of domesti- andAgriculture. http://www.fao.org/giews/. cated species in the diet typically postdate the origin of agri- Weber,M. culture by a millennium or more. 1930 TheProtestant Ethic and the Spiritof Capitalism,trans- 2. It has also been argued that Pleistocene climates were latedby TalcottParsons. Allen Unwin, London. less seasonally variable than during the Holocene, but this Werne,J. P., D. J. Hollander,T. W. Lyons,and L. C. Peterson idea has scant 2000 Climate-InducedVariations in Productivityand Plank- empirical support(Miracle and O'Brien 1998). tonic Ecosystem Structure from the Younger Dryas to Elias (1999) has used fossil beetle faunas to estimate July Holocene in the CariacoBasin, Venezuela. Paleoceanogra- and January temperatures in Holocene and Pleistocene phy 15:19-29. deposits. These data suggest that the Pleistocene was more West,G. J. seasonal than the Holocene. However, beetle estimates of 2000 Pollen Analysis of Late Pleistocene-HoloceneSedi- Januarytemperatures are not very reliable because beetles in ments fromCore CL-73-5, ClearLake, Lake County,Cal- temperate and arctic climates overwinter in a dormant state ifornia:A TerrestrialRecord of California'sCismontane so that their distributions are rather insensitive to winter as Vegetationand Climate Change Inclusive of the Younger opposed to summertemperatures. Plant distributionsare sim- Dryas Event. In Pioceedings of the SeventeenthAnnual ilarly affected. No current method of estimating winter tem- Pacific ClimateWorkshoop, May 22-25, 2000, edited by G. J. West and L. D. Buffaloe, pp. 91-106. InteragencyEco- peraturesin the Pleistocene is reliable. logical Programfor the San FranciscoEstuary Technical 3. Agronomists and climatologists have recently become Report67. CaliforniaDepartment of WaterResources, Sacra- interested in the impacts of climate change and climate vari- mento. ability in the context of CO2-induced global warming Winterhalder,B., and C. Goland (Bazzaz and Sombroek 1996; Downing et al. 1999; Kane and 1997 An EvolutionaryEcology Perspectiveon Diet Choice, Yohe 2000; Reilly and Schimmelpfennig 2000; Rosensweig Risk,and Plant Domestication. In People, Plants, and Land- and Hillel 1998; Schneider et al. 2000). Global climate mod- scapes: Studiesin Paleoethnobotany,edited by K. J. Gremil- els suggest that global warming may increase short time- lion, pp. 123-160. Universityof AlabamaPress, Tuscaloosa. Winterhalder,B., and F. l u scale climate variation as well as creating a steep trend. To 1995 A Forager-ResourcePopulation Ecology Model and some degree, these conditions mimic the millennial and sub- Implicationsfor Indigenous Conservation.Conservation millennial scale variations in the Pleistocene, and, as crop- Biology I1: 1354-1364. and-weather models and empirical data improve, more Wohlgemuth,E. definitive assessments of impact of last glacial conditions on 1996 ResourceIntensification In PrehistoricCentral Califor- plant-based subsistence strategies will become possible. nia:Evidence From Archaeobotanical Data. Journal of Cal- 4. By "adaptive"we mean behaviors that, by comparison iforniaand GreatBasin Archaeology 18:81-103. with available alternatives, have the largest population mean Wright,H.E., Jr. fitness. 1977 EnvironmentalChange and the Originof Agriculturein 5. Some human populations might have curtailed birth the Old and New Worlds.In Originsof Agricuhlture,edited by C. A. Reed, pp. 281-318. Mouton,The Hague. rates in order to preserve higher incomes at any given level of Zvelebil, M. intensification. In a sense, such populations have just rede- 1996 The AgriculturalFrontier and the Transitionto Farm- fined K to be a lower value that permits higher incomes by ing in the Circum-BalticRegion. In ThleOrigins and Spread employing what Malthus called the "preventativechecks" on of Agricultureand Pastoralisinin Eurasia,edited by D. R. population growth. The riest of the above analysis then Harris,pp. 323-345. SmithsonianInstitution Press, Wash- applies with K measuredin suitably emic terms. Culturaldif- ington,D.C. ferences in the value of intensification threshold or K (Coale 1986) will make evidence of stress more likely in populations Notes where the effective carryingcapacity is closer to the ultimate 1. We define "efficiency"as the productivityper unit area subsistence carrying capacity than in populations that reduce of land exploited for subsistence. Efficiency of subsistence is growth rates by preventative checks that keep population favored by strategies that move subsistence down the food well below absolute subsistence limits. The perceived costs chain, especially to high-productivity plant riesources. of populationcontrol, given that the main mechanism in non- "Intensification"we define as the use of human labor to add modern societies was infanticide and sexual abstinence, may productive lower-rankedresources to the diet or the use of mean that most populations intensified labor inputs at any technological innovations to increase the rank of more pro- given level of technology efficiency to near subsistence lim- ductive resour-ces.Typically both strategies are employed its (Hayden 1981). In either event, population pressure will simultaneously.Since increases in efficiency are achieved by tend to stay constant to the extent that rates of population Richerson et al.] AGRICULTURALORIGINS AND CLIMATE 411 growth and intensification are successful in adjusting subsis- nerableto competitive displacement by the intensifiers unless tence to currentconditions. Normally population growth and the greater wealth of the population limiters allows them to decline are quite rapid processes relative to rates of innova- successfully defend their resource-rich teririitories. On the tion and will keep average population size quite close to K. evidence of the fairly rapid rate of spread of intensified Short-termdepartures from K caused by short-termenviron- strategies once invented, such defense is seldom successful mental shocks and windfalls should be the commonest rea- (e.g., Ammerman and Cavalli-Sforza 1984; Bettinger and sons to see especially stressed or unstressed populations. If Baumhoff 1982). the rate of innovation is more rapid than exponential popula- 6. The dates in Table 1 reflect considerable recent revi- tion growth for any significant time period, then per capita sion stemming from accelerator mass spectrometry '4C dat- incomes can rise under a regime of very rapid population ing, which permits the use of very small carbon samples and growth, as in the last few centuries. This regime, if it had can be applied directly to carbonized seeds and other plant occurred in the past, should be quite visible in the historical parts showing morphological changes associated with and archaeological record because it so rapidly leads to large domestication. Isolated seeds tend to work their way deep populations and large-scale creation of durable artifacts. into archaeological deposits, and dates based on associated Alternatively, population growth may have been limited in large carbon samples (usually charcoal) often gave anom- past populations by the analog of the modern demographic alously early dates. transition. Thus, hunter-gatherersmight have resisted the adoption of plant-based intensification because they viewed the life style associated with plant collecting or planting as a decrement to their incomes. However, resisting intensifica- Received February2, 2000; Revised July 21, 2000; tions that increase human densities makes such groups vul- Accepted September 15, 2000.